Yes, quite right. There's nothing magic about the way that the chip works. The problem is more a lack of appropriate terminology. Because the signal is not recorded or stored in the chip it doesn't always make sense to talk about playing it back "slowed down" or "backwards". The original phrase "time dilation" was coined many years ago to try and describe the reason why the signal is so slow. This was even more confusing so it was dropped in favor of "time expansion". I take your point that it's time to drop this moniker as well and find a less sensational term.
This kind of feedback is critical to making the technology acceptable in the minds of potential users. I think that most engineers (myself included) get too lost in the subtleties of their designs and drift far away from potential customers. Before this chip becomes a success, there are more than just technical hurdles to overcome. The device needs to sound plausible and easy to use. I thank you for raising the issue.
This "slowing down" of the signal is the biggest benefit AND the biggest compromise of the design. Because you are not dealing with the real-time signal, you have to wait for the slowed down signal to appear before you can time it. It means that very high speed measurement rates are not possible.
Almost every sentence you write has some technical merit but they all are not entirely correct... It’d be too much work for me to go over each one, sorry…
"One of the reasons why I am presenting my ideas on this forum is to get feedback ".
That's what I thought - just to attract the audience. Hence the Time Machine in the title too, right? I like your location too. Good work! I'm out...
For less than 20m, ultrasonic can be used. PING goes to 3m, so there might be a blind spot unless other solutions are used (higher power, different frequency, more expensive transducer etc). If the part has range 20-156m, that is fine. I can find a use for it.
The use of ultrasonics for short range detection may not be quite so clear cut. The wavelengths used by ultrasonic devices makes them susceptible to "reflections", or more correctly, "echoes". This can lead to signals failing to come back (being reflected away) or in some situations, multiple reflections.
In contrast, at the wavelengths of light the laws of diffuse reflection apply and there is almost always a signal that comes straight back to the receiver. Many of the industrial designs that I have done were used over a range of less than 5m for this very reason.
Without wanting to risk the wrath of alex123, the explanation of the differences between specular and diffuse reflections can best be found on the Internet.
It means that very high speed measurement rates are not possible.
Thanks for clearing that up. I now feel I understand how it works, and that I and others will be able to use this in applications.
I didn't do the math, but I don't think that very high rates of measurement are needed. ANY rate is useful. Even if dozens or hundreds of samples are averaged to increase accuracy, something would have to be moving fairly fast to do not appear standing still compared tothe speed of light. And if something WERE moving very fast, then accuracy would not be so important (since its already moved) and lower accuracy at few samples is fine.
Most hobbyist robot projects move fairly slowly anyway, due to the crashing during testing thing, so everything looks good so far. If its light enough to put on a quadcopter, we can start talking about ranges and data rates at a given speed, altitude etc. For now I'm more focused more along the lines of a zhu-zhu pet.
industrial designs that I have done were used over a range of less than 5m for this very reason.
Does this mean that your design could also be used less than 20m or even less than 5m? If this were possible, this would be very useful.
In addition to finding things far away for "position in the world" data, robots also want to find "position right next to me" data for finding an manipulating objects. For example, the robo-magellan competition asks us to navigate to a series of orange safety cones, and tap each once. The robots can use GPS to get near the cones, but then can have a hard time finding the actual cone without expensive sensors. If we could determine that is an object about this far away, and its about this wide and this high, that would be very helpful.
I'll read up on specular and diffuse reflections. I'm sure alex123 will happily return when there are parts to play with.
So speaking of parts, do you need anymore information?
From the time the information is collected, about how long do you estimate till a usable prototype is ready? Are we talking months or years?
Do you still think you will be able to produce a part tthat sells for $20?
...This was even more confusing so it was dropped in favor of "time expansion". I take your point that it's time to drop this moniker as well and find a less sensational term.
Perhaps it's time to start getting used to more "sensational terms"... I expect we'll start seeing many more of them soon, and they'll be quite real!
I'm thoroughly fascinated by this thread, I wish I had something more to contribute, but I wanted to at least thank you for all the terms and info which have given me some fascinating material to study the past few hours!
Thanks for the comment xantos. It's hard to keep a thread like this going when I have to think clearly at one in the morning. I really am located in deepest, darkest Africa.
I wasn't clear on the range issue. 20m is the shortest "maximum range" that can be set. The longest "maximum range" is 180m. In both cases the minimum measuring range is zero. Laser distance measurement for close detection works very well.
The price of the part is determined by the technology we use, not by how simple or complex we make the design. This is because certain features require certain types of fabrication and the LVPECL interface turns out to be the structure that is adding the most cost. That being said, the $20 price tag will easily be achieved.
As for the time frame, once we've established all the operating parameters it will be days or weeks not months or years to produce the chip. It will take longer to make the circuit boards. The chip itself is currently in a 100 pin quad flat pack (same as my avatar picture).
For those who like to see some hardware, I've attached a badly taken picture of one my earlier rangefinders.
I have been worrying about the use of appropriate terminology particularly since I am trying to write the datasheet for the DS00VQ100 chip. It seems to me that the ideas need to be presented in familiar terms so that users can understand how things work without being bogged down in techno-jargon. I have a suggestion about how to explain the concept of "slowing down" a signal.
We are all familiar with amplifiers. Amplifiers take signals and make them bigger. In other words, they increase the "amplitude" of the signal on a voltage scale. Some amplifiers can also invert the signal to make it look like it is upside down. Operational amplifiers need "feedback" to limit the gain otherwise they become unusable. Amplifiers don't store the signal, they act on it as it comes in and immediately change its characteristics.
The method of slowing down the signal in the rangefinder chip is closely analogous to an amplifier but instead of acting on the amplitude of the signal, it acts on the timebase of the signal. Like an amplifier, you can adjust its gain to make the signal slow down a bit or a lot. Like an amplifier you can choose to invert the timebase. Like an amplifier, it needs feedback to prevent very high gain or instability. Finally, like an amplifier, it works on the live signal as it comes in, not on a stored image.
Perhaps the term "time amplifier" may be more acceptable?
- timebase amplification?
- timebase gain?
I don't think "time amplifier" will work.. people will just ask 'how can you amplify time?' Most people (including digital techheads) are not really aware of the feedback circuits in an amplifier and other subtle (and interesting) aspects.
"time domain moderator" maybe? Hm, I'm not too happy with 'moderator'.. but English not being my native language I've trouble recalling the term that I'm really thinking of. Or maybe even "time domain amplifier" may be working better than just "time amplifier".. but there's another word that keeps eluding me. I can see it in the corner of my eye but when I look it moves away..
...appropriate terminology... suggestion about how to explain the concept of "slowing down" a signal.
timebase. .... it works on the live signal as it comes in, not on a stored image.
Perhaps the term "time amplifier" may be more acceptable?
- timebase amplification?
- timebase gain?
Unless I'm tottally missing it, time remains constant. Time is not stretched, expanded, amplified, or anything aside from passing normally. Hinting otherwise is what turned off alex123 and the lurkers that would normally be interested but did not post. The line will not sell in these parts.
As I see it, it does in fact work on a stored signal in this fashion: The part know the start of the signal, and detects the return signal. The "know the start of the signal" is the stored part. AFTER the return is recieved (again, consider this as stored) the part does whatever to increase the scaling of the difference between start and end by a factor of one million for exmple, and determines the distance for this. There is error in this, so it must take multiple reading, and averages out the error.
If this is correct, then this might be better described as a "frequency down-shifter" or "signal expander".
If you choose to try to put it in terms of timebase, please be very cautious. You might have an acceptable explanation is if you use an analogy of changing the tempo of a musical peice from "Allegro" to "Adagio", as in 130 beats per minute to 10 beats per minute.
I strongly advise against hinting that you change the timebase of time itself, which was the impression I got at first. Most folks here would dismiss the effort as BS regardless of the capability of the part.
"Signal stretcher" is a term you might want to consder to describe the technic you are using for your laser time of flight range finder.
Very interesting that this is such a complicated point.
I've reworded the datasheet to use "resolution" and it reads just fine in the sense that you can still get the chip to do what you want even if the "signal stretching" is happening in the background. It would be nice to give an interesting name to the novel feature but maybe it's just better to avoid it for now. There are other things that still need to be done...
Up until now we have been discussing the chip itself. The next step is to look at the physical constrains of the final design. Often this process helps to rationalize features or uncovers missing elements.
The laser rangefinder consists of four parts:-
1. The laser with its associated electronics and optics.
2. The receiver with its associated electronics and optics.
3. A controlling and timing system.
4. An interface to a micro or other master device.
In my other designs these parts were all jammed into a single housing. This is fine if you have a single application in mind but is very limiting if you want to experiment and explore other uses. I would like to consider the idea of making modules that can be used in different ways.
For example, a small system with a range of a few meters only requires small optical components whereas an autonomous vehicle may need a range of a 100m but is big enough to carry much larger optics. There is no reason why the same laser, receiver and timing modules shouldn't work in both applications. The only difference would be in the design of the optical parts.
What is important for the chip design is to understand how these parts connect together electronically so that the right number and types of signals can be conveyed between the modules.
Alternatively, we can select a single application and build a "once-off" design.
I don't mean to be raining on the parade, but that is what I was trying to get at in the "cute" comment in #24. It is not complicated, but it is very important.
The explanation in sentence 1 of post #1 interesting enough to pique our interest. However, "Time Machine" is implies something that is obviously not to be taken at face value, as it implies an H.G. Wells story concept. Unless yo can show that H.G. Wells fantasy is now reality, you better avoid that term, or draw the wrath of alex123 etc. "Time Expander" is equally dangerous. Unless you can demonstrate that time is to be expanded, a more on the mark term should be used. Same for timebase, although this is not so dangerous, but have should resist your tendency implying that you are doing sometimes to time.
In reality, the brief signal is being scaled up by a novel means to allow measurement using conventional means. This is a lily that doesn't need gilding.
A working part will yield many sales at $20 a pop. I don't think I've seen anything like this, or I would have one.
OK, this topic has been beaten to death. I will once again attempt to refrain.
The laser rangefinder consists of four parts:-
1. The laser with its associated electronics and optics.
2. The receiver with its associated electronics and optics.
3. A controlling and timing system.
4. An interface to a micro or other master device.
noted
The only difference would be in the design of the optical parts.
noted
Alternatively, we can select a single application and build a "once-off" design.
If the design can accommodate
1. a generic laser diode as found in a laser pointer
2. a prop (80Mhx, SPI interface)
3. Optics per application (with some means of the user deterimins what optics are appropriate and how to get them)
4. a "standard detector" most folks (me in particular) don't have a wide experience with laser detectors, one should be included or at least specified
I like the idea of a kit. Perhaps we could offer two projects in one - a short range and a long range with different optics.
1. The laser has to be a high power, pulsed type. We could give a reference design for those who want to make their own and offer a complete module for those who don't. There are many suppliers including companies like RS Components. I would suggest the Osram parts because they are easy to use:
SPL PL90 (75 Watt) for a long range project.
SPL LL85 (14 Watt) for a short range project.
Osram also has great reference information on their website.
2. Is there a particular form factor (shape and pinouts) that would work with the Prop?
3. Plastic optics are a good option because they are very cheap but here is where performance can really be optimized. We'll have to decide on realistic sizes for a kit and let the enthusiasts make their own if they want something special.
4. Detectors can be either standard PIN diodes or more sophisticated APD's (avalanche photo diodes). Like with the lasers we could offer a reference design or a complete module for both. There is one critical trick though - it is important to have a narrow band, optical filter in front of the detector if the project is to be used outside. This keeps out background light. Performance will even be improved if red plexiglass or similar is put in front of the receiver.
It would be interesting to hear from people who know of a low cost source of supply for these kinds of parts. I've got some suppliers for the more specialized, high performance parts but they might be too expensive for the hobbyist.
Don't you need special something if its above some milliWatt limit? This page http://en.wikipedia.org/wiki/Laser_safety
makes me think the target should be Class 3R (5 mW) or less. Can the low power be compensated for with better optics, etc?
Otherwise, how can we put it on a robot that is intended for the outsiode of a lab?
2. Is there a particular form factor (shape and pinouts) that would work with the Prop?
The prop bread boards and most board kits have 0.01 pins. Ususally 3.3v and ground are on adajenct pins, (schmart board, proto board) but not always and in the same orientation. I usually get an old IDE cable, and put a row of pin headers on the end. I clip the pair on the end for power and ground (so they can be swapped for different boards) and use consecutive pint in groups of 0-7, 8-15, 16-23, and 24-27. NOTE that Pin 28,29, 30 and 31 are usually reserved for EEPROM and serial terminal, this is not manadatory, but it is convenient and I've never run out of pins and needed to reruse these.
The physical form factor of the boards is not "standard" even amoung parallax offerings. It is safe to assume that the prop board would be any and all of the parallax boards and third party boards. It is also safe to assume that the board used in experiments need not be the board used in a final design, robot or otherwise, as a custom layout is called for in many cases.
3. Plastic optics are a good option because they are very cheap but here is where performance can really be optimized. We'll have to decide on realistic sizes for a kit and let the enthusiasts make their own if they want something special.
Excellent idea. My rule of thumb is "minimum necessary and sufficient". In this case I would advise "minimum necessary and sufficient to measure a distance in the lab." Once I get it working on the bench, I can decide what to do to use it in a given situation. Therefore, cheapest and easiest are appropriate, since the intent is that the use will a custom solution based on individual need.
4. Detectors ...PIN diodes or ....APD's
one critical trick ... narrow band, optical filter
Reference design, and choice of "minimum necessary and sufficient for the lab" versus "preferred APD" would be great. This would allow use to control cost based on need. Also a separate APD upgrade?
low cost source of supply for these kinds of parts.
Nutty hobbyist will spend whatever it takes to get the project done, but more general users prefer cheap. Many have bench supplies, I actually have a 50volt 18 amp rack mountable power supply that I am afraid to turn on. Most use the wall wart (7.2V 1.2A) supplied by parallax as it come with the kits.
If the rig can use 4 AA LiPO, this would be idea for a portable application.
A mobile robot could be assumed to have larger batteries in multiples of 12volts, but the 4.8v would be a more general solution.
...
So, the big question seem to be how to deal with the 14 Watt and 75 Watt laser complications. Is there an option to use 3 mW laser and better optics; or use a very short pulse to eliminate the need for eye protection or something?
A 75W laser may sound like a monster but it's really a sheep in wolf's clothing. The eye safety regulations take into account peak power, average power and energy. It is quite easy to make a design "class 1 eye safe".
For example, suppose that a "lab" design used a 14W laser running at 10kHz with a pulse width of 10ns. The average power would only be 1.4mW. However, eye safety is an important issue and we must use the appropriate warnings. In practice, these pulsed lasers are usually far safer than their CW counterparts.
I will check on the design options for the Prop boards as per your description. The LRF shown in the picture in #39 is an example of "form factor" gone crazy. It uses a 7W pulsed laser and an analog timing system.
Whilst some will disagree, making a receiver module with an APD is actually easier than using a PIN diode. The reason is that the APD has hundreds of times more sensitivity than a PIN so the post-amplification is much easier. That means cheaper parts and better performance. Also, don't underestimate the fun factor. It's cool to wind an APD up to its maximum gain and find a weak target (return signal) buried in the noise. And then try to get the distance...
I think that the idea of making the system run on 4.8V is fantastic! I've never tried this before but the math suggests that it is possible.
75W laser may sound like a monster but it's really a sheep in wolf's clothing..... It is quite easy to make a design "class 1 eye safe".
Excellent! A 75W laser is so much cooler anyway.
... average power would only be 1.4mW. ... pulsed lasers are usually far safer than their CW counterparts.
I never would have guessed. Do you have to ensure that a maximum pulse is limited, or is this in the nature of the chip? Some nut (like me) might try to hack it into continuous mode or other dangerously stupid mod, and there could be liability issue?
..."form factor" gone crazy.
I would think the cheapest, smallest board to do the job would be fine. If that board happens to be some other standard, its a bonus, but not mandatory.
... That means cheaper parts and better performance.
You seem to be selecting the most cost effective options, I can have no objections unless somebody can offer valid technical reasons against your design choises.
This sounds like it shaping up to be a kickin' rig. But now I'm starting to question the $20 target price. Did the $20 refer to just the chip, or to the whole rig? What can we expect if we include the 7W/75W pulsed laser and electionics, and the minimum optics, and the avalanche photo diode detector and electronics? It would be amazing if any of these could be included in the chip itself.
Even with using a APD you would probably want a large lens to collect the returning light and focus it onto the sensor. I was thinking using standard camera parts with a IR filter would work pretty well. Light would be filtered so only IR gets in then collected by the lets say 2inch lens and then focused onto the APD. This would help with the sensitivity and the range.
I gotta know, are you at Lake Tanganyika in Ujiji?
Nice place, but no. I'm North of Johannesburg in an area called Midstream Estate. Looking out of my window the fields are burnt black and ash is lying everywhere. The setting sun is bleeding red into the night sky but the land is as monochrome today as it will be under the glare of the moon tonight. I find that Africa is a place that torments the soul but offers the most enlivening experiences.
Even with using a APD you would probably want a large lens to collect the returning light and focus it onto the sensor. I was thinking using standard camera parts with a IR filter would work pretty well. Light would be filtered so only IR gets in then collected by the lets say 2inch lens and then focused onto the APD. This would help with the sensitivity and the range.
Your calculation is very close. For short range applications of less than 20m you can get away with between 1/2 and 1 inch. But you might be surprised to know that a 2 inch lens of camera quality, maybe 4 inch focal length, can measure over 500m to a reasonably co-operative target (white) and several miles to a reflective target.
FAQ - What quality of lens is required for the receiver in a laser ranger finder?
ANS - It depends. The reason is that several effects come into play when measuring objects that are close.
The first is parallax. Not the micro kind, but the optical kind, caused by the image of the laser beam moving off-axis as the target gets closer. A perfectly aligned optical system with good quality lenses will work well at long range but will lose performance at short range. This is especially true for long focal length lenses which give spectacular results over long range and zero signal strength when close up.
Another effect is focus. Because the detector is placed at the best focal point for long range use it goes out of focus as the target gets nearer. This means that the image of the laser spot gets blurry and loses strength as the target comes closer.
The surprising discovery is that the combined result of these two phenomena means that the performance of a rangefinder gets better as the target moves further away. This is of course true only up to a point.
For those with optical design experience, I'm sure that you can come up with a perfect system that will work for all ranges - beam splitters, concentric lenses, bifocals etc. The real challenge is to design a system that "self-compensates" for the inverse square law loss of signal strength with distance. Ideally, the signal strength close up should be identical to that far away.
For our project I would suggest using lens of particularly poor quality. Lots of spherical and chromatic aberration and a short focal length. That means that your average magnifying lens may well work. I have used a molded acrylic lens of 2" diameter and 3" focal length to measure 200m to a white wall. The signal strength at 50m was identical to the signal strength at 1m.
But now I'm starting to question the $20 target price. Did the $20 refer to just the chip, or to the whole rig? What can we expect if we include the 7W/75W pulsed laser and electionics, and the minimum optics, and the avalanche photo diode detector and electronics? It would be amazing if any of these could be included in the chip itself.
The $20 is just for the chip. It's not possible to include the APD or laser in it but...
One of the things that the chip does have is controllers for the laser and APD. This may not sound like much until you discover how notoriously hard these things are to manage.
Firstly, the high power, pulsed laser. It is easy to talk about 75W of output power put how much electrical power is needed? The answer is about 800W - 40 Amps at 20 V. DON'T PANIC! There is a big difference between instantaneous power and average power. At 10kHz firing rate an ideal power supply will only need to deliver about 8 mW. There are tricks to designing this power supply to make sure that it delivers the exact amount of energy to the laser at precisely the moment that the laser needs it. The chip handles this with a synchronous pulse-width-modulated controller. A comparator, a small inductor and a transistor will complete the circuit (diagrams later).
Secondly, the APD. This is a most non-linear, temperature sensitive and anoying device to control. One moment you are running at a gain of 1000 and getting a fantastic signal, the next it's spewing noise into your receiver. Many vendors offer complete control modules for their APD's. These modules have temperature sensors and current controls to get some stability on the gain. Oh, and I didn't menion that the kind of APD's that we will use may need more than 200V DC to operate. APD modules are very expensive. Fortunately, our trusty chip comes to the rescue again. It doesn't try to emulate other designs, rather, it takes a very practical approach to the problem. The chip monitors the background noise on the return signal using a separate, internal circuit. If the noise increases then either the temperature has changed or the background light has increased. Either way the gain needs to be dropped to keep the noise out of the measurement. The APD controller manages this process. The chip has another, synchronous power supply for the APD. Once again we need a small inductor, a comparator and a transistor and the APD is tamed (diagrams later).
So whilst there will be a few expensive parts - the chip, the laser, the APD and the Propeller (?), the rest of the system is inexpensive.
Comments
This kind of feedback is critical to making the technology acceptable in the minds of potential users. I think that most engineers (myself included) get too lost in the subtleties of their designs and drift far away from potential customers. Before this chip becomes a success, there are more than just technical hurdles to overcome. The device needs to sound plausible and easy to use. I thank you for raising the issue.
This "slowing down" of the signal is the biggest benefit AND the biggest compromise of the design. Because you are not dealing with the real-time signal, you have to wait for the slowed down signal to appear before you can time it. It means that very high speed measurement rates are not possible.
"One of the reasons why I am presenting my ideas on this forum is to get feedback ".
That's what I thought - just to attract the audience. Hence the Time Machine in the title too, right? I like your location too. Good work! I'm out...
The use of ultrasonics for short range detection may not be quite so clear cut. The wavelengths used by ultrasonic devices makes them susceptible to "reflections", or more correctly, "echoes". This can lead to signals failing to come back (being reflected away) or in some situations, multiple reflections.
In contrast, at the wavelengths of light the laws of diffuse reflection apply and there is almost always a signal that comes straight back to the receiver. Many of the industrial designs that I have done were used over a range of less than 5m for this very reason.
Without wanting to risk the wrath of alex123, the explanation of the differences between specular and diffuse reflections can best be found on the Internet.
Thanks for clearing that up. I now feel I understand how it works, and that I and others will be able to use this in applications.
I didn't do the math, but I don't think that very high rates of measurement are needed. ANY rate is useful. Even if dozens or hundreds of samples are averaged to increase accuracy, something would have to be moving fairly fast to do not appear standing still compared tothe speed of light. And if something WERE moving very fast, then accuracy would not be so important (since its already moved) and lower accuracy at few samples is fine.
Most hobbyist robot projects move fairly slowly anyway, due to the crashing during testing thing, so everything looks good so far. If its light enough to put on a quadcopter, we can start talking about ranges and data rates at a given speed, altitude etc. For now I'm more focused more along the lines of a zhu-zhu pet.
Does this mean that your design could also be used less than 20m or even less than 5m? If this were possible, this would be very useful.
In addition to finding things far away for "position in the world" data, robots also want to find "position right next to me" data for finding an manipulating objects. For example, the robo-magellan competition asks us to navigate to a series of orange safety cones, and tap each once. The robots can use GPS to get near the cones, but then can have a hard time finding the actual cone without expensive sensors. If we could determine that is an object about this far away, and its about this wide and this high, that would be very helpful.
I'll read up on specular and diffuse reflections. I'm sure alex123 will happily return when there are parts to play with.
So speaking of parts, do you need anymore information?
From the time the information is collected, about how long do you estimate till a usable prototype is ready? Are we talking months or years?
Do you still think you will be able to produce a part tthat sells for $20?
Perhaps it's time to start getting used to more "sensational terms"... I expect we'll start seeing many more of them soon, and they'll be quite real!
I'm thoroughly fascinated by this thread, I wish I had something more to contribute, but I wanted to at least thank you for all the terms and info which have given me some fascinating material to study the past few hours!
Dave
The price of the part is determined by the technology we use, not by how simple or complex we make the design. This is because certain features require certain types of fabrication and the LVPECL interface turns out to be the structure that is adding the most cost. That being said, the $20 price tag will easily be achieved.
As for the time frame, once we've established all the operating parameters it will be days or weeks not months or years to produce the chip. It will take longer to make the circuit boards. The chip itself is currently in a 100 pin quad flat pack (same as my avatar picture).
For those who like to see some hardware, I've attached a badly taken picture of one my earlier rangefinders.
We are all familiar with amplifiers. Amplifiers take signals and make them bigger. In other words, they increase the "amplitude" of the signal on a voltage scale. Some amplifiers can also invert the signal to make it look like it is upside down. Operational amplifiers need "feedback" to limit the gain otherwise they become unusable. Amplifiers don't store the signal, they act on it as it comes in and immediately change its characteristics.
The method of slowing down the signal in the rangefinder chip is closely analogous to an amplifier but instead of acting on the amplitude of the signal, it acts on the timebase of the signal. Like an amplifier, you can adjust its gain to make the signal slow down a bit or a lot. Like an amplifier you can choose to invert the timebase. Like an amplifier, it needs feedback to prevent very high gain or instability. Finally, like an amplifier, it works on the live signal as it comes in, not on a stored image.
Perhaps the term "time amplifier" may be more acceptable?
- timebase amplification?
- timebase gain?
"time domain moderator" maybe? Hm, I'm not too happy with 'moderator'.. but English not being my native language I've trouble recalling the term that I'm really thinking of. Or maybe even "time domain amplifier" may be working better than just "time amplifier".. but there's another word that keeps eluding me. I can see it in the corner of my eye but when I look it moves away..
-Tor
We could state - "adjustable range and adjustable resolution"
Unless I'm tottally missing it, time remains constant. Time is not stretched, expanded, amplified, or anything aside from passing normally. Hinting otherwise is what turned off alex123 and the lurkers that would normally be interested but did not post. The line will not sell in these parts.
As I see it, it does in fact work on a stored signal in this fashion: The part know the start of the signal, and detects the return signal. The "know the start of the signal" is the stored part. AFTER the return is recieved (again, consider this as stored) the part does whatever to increase the scaling of the difference between start and end by a factor of one million for exmple, and determines the distance for this. There is error in this, so it must take multiple reading, and averages out the error.
If this is correct, then this might be better described as a "frequency down-shifter" or "signal expander".
If you choose to try to put it in terms of timebase, please be very cautious. You might have an acceptable explanation is if you use an analogy of changing the tempo of a musical peice from "Allegro" to "Adagio", as in 130 beats per minute to 10 beats per minute.
I strongly advise against hinting that you change the timebase of time itself, which was the impression I got at first. Most folks here would dismiss the effort as BS regardless of the capability of the part.
"Signal stretcher" is a term you might want to consder to describe the technic you are using for your laser time of flight range finder.
You do want to say this, but you still want to describe the technique as "signal stretcher" as this is the novel part.
I've reworded the datasheet to use "resolution" and it reads just fine in the sense that you can still get the chip to do what you want even if the "signal stretching" is happening in the background. It would be nice to give an interesting name to the novel feature but maybe it's just better to avoid it for now. There are other things that still need to be done...
The laser rangefinder consists of four parts:-
1. The laser with its associated electronics and optics.
2. The receiver with its associated electronics and optics.
3. A controlling and timing system.
4. An interface to a micro or other master device.
In my other designs these parts were all jammed into a single housing. This is fine if you have a single application in mind but is very limiting if you want to experiment and explore other uses. I would like to consider the idea of making modules that can be used in different ways.
For example, a small system with a range of a few meters only requires small optical components whereas an autonomous vehicle may need a range of a 100m but is big enough to carry much larger optics. There is no reason why the same laser, receiver and timing modules shouldn't work in both applications. The only difference would be in the design of the optical parts.
What is important for the chip design is to understand how these parts connect together electronically so that the right number and types of signals can be conveyed between the modules.
Alternatively, we can select a single application and build a "once-off" design.
The explanation in sentence 1 of post #1 interesting enough to pique our interest. However, "Time Machine" is implies something that is obviously not to be taken at face value, as it implies an H.G. Wells story concept. Unless yo can show that H.G. Wells fantasy is now reality, you better avoid that term, or draw the wrath of alex123 etc. "Time Expander" is equally dangerous. Unless you can demonstrate that time is to be expanded, a more on the mark term should be used. Same for timebase, although this is not so dangerous, but have should resist your tendency implying that you are doing sometimes to time.
In reality, the brief signal is being scaled up by a novel means to allow measurement using conventional means. This is a lily that doesn't need gilding.
A working part will yield many sales at $20 a pop. I don't think I've seen anything like this, or I would have one.
OK, this topic has been beaten to death. I will once again attempt to refrain.
If the design can accommodate
1. a generic laser diode as found in a laser pointer
2. a prop (80Mhx, SPI interface)
3. Optics per application (with some means of the user deterimins what optics are appropriate and how to get them)
4. a "standard detector" most folks (me in particular) don't have a wide experience with laser detectors, one should be included or at least specified
this could be an appealing hobbyist kit.
1. The laser has to be a high power, pulsed type. We could give a reference design for those who want to make their own and offer a complete module for those who don't. There are many suppliers including companies like RS Components. I would suggest the Osram parts because they are easy to use:
SPL PL90 (75 Watt) for a long range project.
SPL LL85 (14 Watt) for a short range project.
Osram also has great reference information on their website.
2. Is there a particular form factor (shape and pinouts) that would work with the Prop?
3. Plastic optics are a good option because they are very cheap but here is where performance can really be optimized. We'll have to decide on realistic sizes for a kit and let the enthusiasts make their own if they want something special.
4. Detectors can be either standard PIN diodes or more sophisticated APD's (avalanche photo diodes). Like with the lasers we could offer a reference design or a complete module for both. There is one critical trick though - it is important to have a narrow band, optical filter in front of the detector if the project is to be used outside. This keeps out background light. Performance will even be improved if red plexiglass or similar is put in front of the receiver.
It would be interesting to hear from people who know of a low cost source of supply for these kinds of parts. I've got some suppliers for the more specialized, high performance parts but they might be too expensive for the hobbyist.
interchangable optics is a great idea.
excellent.. wait a minute...
Don't you need special something if its above some milliWatt limit? This page
http://en.wikipedia.org/wiki/Laser_safety
makes me think the target should be Class 3R (5 mW) or less. Can the low power be compensated for with better optics, etc?
Otherwise, how can we put it on a robot that is intended for the outsiode of a lab?
The prop bread boards and most board kits have 0.01 pins. Ususally 3.3v and ground are on adajenct pins, (schmart board, proto board) but not always and in the same orientation. I usually get an old IDE cable, and put a row of pin headers on the end. I clip the pair on the end for power and ground (so they can be swapped for different boards) and use consecutive pint in groups of 0-7, 8-15, 16-23, and 24-27. NOTE that Pin 28,29, 30 and 31 are usually reserved for EEPROM and serial terminal, this is not manadatory, but it is convenient and I've never run out of pins and needed to reruse these.
The physical form factor of the boards is not "standard" even amoung parallax offerings. It is safe to assume that the prop board would be any and all of the parallax boards and third party boards. It is also safe to assume that the board used in experiments need not be the board used in a final design, robot or otherwise, as a custom layout is called for in many cases.
Excellent idea. My rule of thumb is "minimum necessary and sufficient". In this case I would advise "minimum necessary and sufficient to measure a distance in the lab." Once I get it working on the bench, I can decide what to do to use it in a given situation. Therefore, cheapest and easiest are appropriate, since the intent is that the use will a custom solution based on individual need.
Reference design, and choice of "minimum necessary and sufficient for the lab" versus "preferred APD" would be great. This would allow use to control cost based on need. Also a separate APD upgrade?
Nutty hobbyist will spend whatever it takes to get the project done, but more general users prefer cheap. Many have bench supplies, I actually have a 50volt 18 amp rack mountable power supply that I am afraid to turn on. Most use the wall wart (7.2V 1.2A) supplied by parallax as it come with the kits.
If the rig can use 4 AA LiPO, this would be idea for a portable application.
A mobile robot could be assumed to have larger batteries in multiples of 12volts, but the 4.8v would be a more general solution.
...
So, the big question seem to be how to deal with the 14 Watt and 75 Watt laser complications. Is there an option to use 3 mW laser and better optics; or use a very short pulse to eliminate the need for eye protection or something?
For example, suppose that a "lab" design used a 14W laser running at 10kHz with a pulse width of 10ns. The average power would only be 1.4mW. However, eye safety is an important issue and we must use the appropriate warnings. In practice, these pulsed lasers are usually far safer than their CW counterparts.
I will check on the design options for the Prop boards as per your description. The LRF shown in the picture in #39 is an example of "form factor" gone crazy. It uses a 7W pulsed laser and an analog timing system.
Whilst some will disagree, making a receiver module with an APD is actually easier than using a PIN diode. The reason is that the APD has hundreds of times more sensitivity than a PIN so the post-amplification is much easier. That means cheaper parts and better performance. Also, don't underestimate the fun factor. It's cool to wind an APD up to its maximum gain and find a weak target (return signal) buried in the noise. And then try to get the distance...
I think that the idea of making the system run on 4.8V is fantastic! I've never tried this before but the math suggests that it is possible.
Excellent! A 75W laser is so much cooler anyway.
I never would have guessed. Do you have to ensure that a maximum pulse is limited, or is this in the nature of the chip? Some nut (like me) might try to hack it into continuous mode or other dangerously stupid mod, and there could be liability issue?
I would think the cheapest, smallest board to do the job would be fine. If that board happens to be some other standard, its a bonus, but not mandatory.
You seem to be selecting the most cost effective options, I can have no objections unless somebody can offer valid technical reasons against your design choises.
This sounds like it shaping up to be a kickin' rig. But now I'm starting to question the $20 target price. Did the $20 refer to just the chip, or to the whole rig? What can we expect if we include the 7W/75W pulsed laser and electionics, and the minimum optics, and the avalanche photo diode detector and electronics? It would be amazing if any of these could be included in the chip itself.
I gotta know, are you at Lake Tanganyika in Ujiji?
Could the lens be harvested form a regular handheld magnifying glass?
Nice place, but no. I'm North of Johannesburg in an area called Midstream Estate. Looking out of my window the fields are burnt black and ash is lying everywhere. The setting sun is bleeding red into the night sky but the land is as monochrome today as it will be under the glare of the moon tonight. I find that Africa is a place that torments the soul but offers the most enlivening experiences.
Your calculation is very close. For short range applications of less than 20m you can get away with between 1/2 and 1 inch. But you might be surprised to know that a 2 inch lens of camera quality, maybe 4 inch focal length, can measure over 500m to a reasonably co-operative target (white) and several miles to a reflective target.
FAQ - What quality of lens is required for the receiver in a laser ranger finder?
ANS - It depends. The reason is that several effects come into play when measuring objects that are close.
The first is parallax. Not the micro kind, but the optical kind, caused by the image of the laser beam moving off-axis as the target gets closer. A perfectly aligned optical system with good quality lenses will work well at long range but will lose performance at short range. This is especially true for long focal length lenses which give spectacular results over long range and zero signal strength when close up.
Another effect is focus. Because the detector is placed at the best focal point for long range use it goes out of focus as the target gets nearer. This means that the image of the laser spot gets blurry and loses strength as the target comes closer.
The surprising discovery is that the combined result of these two phenomena means that the performance of a rangefinder gets better as the target moves further away. This is of course true only up to a point.
For those with optical design experience, I'm sure that you can come up with a perfect system that will work for all ranges - beam splitters, concentric lenses, bifocals etc. The real challenge is to design a system that "self-compensates" for the inverse square law loss of signal strength with distance. Ideally, the signal strength close up should be identical to that far away.
For our project I would suggest using lens of particularly poor quality. Lots of spherical and chromatic aberration and a short focal length. That means that your average magnifying lens may well work. I have used a molded acrylic lens of 2" diameter and 3" focal length to measure 200m to a white wall. The signal strength at 50m was identical to the signal strength at 1m.
The $20 is just for the chip. It's not possible to include the APD or laser in it but...
One of the things that the chip does have is controllers for the laser and APD. This may not sound like much until you discover how notoriously hard these things are to manage.
Firstly, the high power, pulsed laser. It is easy to talk about 75W of output power put how much electrical power is needed? The answer is about 800W - 40 Amps at 20 V. DON'T PANIC! There is a big difference between instantaneous power and average power. At 10kHz firing rate an ideal power supply will only need to deliver about 8 mW. There are tricks to designing this power supply to make sure that it delivers the exact amount of energy to the laser at precisely the moment that the laser needs it. The chip handles this with a synchronous pulse-width-modulated controller. A comparator, a small inductor and a transistor will complete the circuit (diagrams later).
Secondly, the APD. This is a most non-linear, temperature sensitive and anoying device to control. One moment you are running at a gain of 1000 and getting a fantastic signal, the next it's spewing noise into your receiver. Many vendors offer complete control modules for their APD's. These modules have temperature sensors and current controls to get some stability on the gain. Oh, and I didn't menion that the kind of APD's that we will use may need more than 200V DC to operate. APD modules are very expensive. Fortunately, our trusty chip comes to the rescue again. It doesn't try to emulate other designs, rather, it takes a very practical approach to the problem. The chip monitors the background noise on the return signal using a separate, internal circuit. If the noise increases then either the temperature has changed or the background light has increased. Either way the gain needs to be dropped to keep the noise out of the measurement. The APD controller manages this process. The chip has another, synchronous power supply for the APD. Once again we need a small inductor, a comparator and a transistor and the APD is tamed (diagrams later).
So whilst there will be a few expensive parts - the chip, the laser, the APD and the Propeller (?), the rest of the system is inexpensive.
I found Highveld Techno Park. It seems pretty built up, is the black ash a result of a 75W laser gone awry?