I've been following along from the beginning, Kramer.
The "1000 guaranteed" is the minimum h_FE specified in the datasheet which is·generally the last place anybody looks for advise or guidance on this Forum.· I see how these pesky PDFs·are... getting in the way.·
Wrong things are being "taught" here, but since you're "having fun" then I don't care.
Hi P.J.
I dug out my copy of 'The Art of Electronics' by Horowitz, et al. to see if I was getting off into the weeds. Searching via Google has not really been as helpful as I feel it should be and my library here is quite limited.
Apparently Horowitz is where I got inspired to drive the TIP120 at 100ma. I refer to pages 61-65 (particularly Section 2.02 - Transistor switch)and Appendix G, wherein he discusses misconceptions about h_FE and the benefits of driving transistors into saturation in the context of transistor switching (not amplification). He seems to explain that saturation with excessive base current is better for a transistor used in switching mode.
To quote him directly, "A circuit that depends on a particular value for h_FE is a bad circuit."
I will leave it to him to fully explain why, but transistors are not truly current amplifiers as represented by h_FE. He points out that they are 'transconductance amplifiers' wherein in a collector current is determined by the voltage applied to the Base-Emitter junction. He later goes into detail about factors that bend that linear relationship into a heavy curve by presenting the Ebers-Moll Model as his preferred means of determining gain -- variations in temperature, current output, and voltage output all change the h-FE substantially. And in a 'transistor switch' it is not unusual to see a simple transistor de-rated to an h_FE between 10 to 20 as great amounts of heat, higher voltages, and higher currents are applied. The darlington pair is not immune to such de-rating, but still provides a higher gain.
I do admit I may not have gotten my explanation of voltage drops correct and I am looking to that now. Doing this bench work has been very helpful to pointing out what I am not seeing in the PDF or what I am reading wrong. After all, if I read it wrong, something on the bench shows me that I should read it again or do more research elsewhere.
Incidentally, this is intended to run an inductive device a motor and that excess saturation may also be required to start the motor. The TIP120 tolerates 5 amps in a steady on, but up to 8 amps momentarily. So I am inclined to stick with 100-120ma driving the base.
The opto-isolator configuration seems a little odd to me, but I wasn't sure how to limit the TIP120's Base voltage to 5 volts. I must say that I do wonder if the voltage regulator is at all necessary. It might be much simpler to drive the TIP120 with a higher value resistor being fed by the automotive '12 - 15 volts'. Apparently this will work, and I will try it on the bench at the risk of burning up another 4N33. It certainly seems to better serve the concept of isolation.
Rather than not point out my mistakes 'because I am having fun', please show me what you feel are genuine misconceptions. Just because I am having fun, doesn't give me license to blindly convey misinformation.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Ain't gadetry a wonderful thing?
aka G. Herzog [noparse][[/noparse] 黃鶴 ] in Taiwan
Post Edited (Loopy Byteloose) : 8/3/2010 9:08:32 PM GMT
Hi all,
Well, "The Art of Electronics" was indeed helpful with pointing the way for me to design a powerful DC switch/solid-state relay equivalent. And yes, a power MOSfet is likely to be a good alternative to the power Darlington BJT. I may do some bench tests of a MOSfet at a later date. Still the TIP120 is very easy to buy and works well if you can tolerate the heat.
I have managed to eliminated the need for the 4N33 to be powered by a 5 volt regulator, so there is no need to run tests for a similar circuit with a 3.3volt regulator to provide nearly 120ma to the base of the TIP120. But a 3.3volt regulator should work with the right current limiting resistor value, say 28 ohms at 1 or 2- watt. (I am using 2 watt because the local store doesn't sell 1 watts, besides it gives me more useful low ohm resistors for other bench tests. Why buy 1/2 watt, 1 watt, and 2 watt values for testing, when a set of 2 watts will do it all?)
The design scheme has been intended to allow the 'high side" to go up to 15 VDC (which is safely beyond the normal limits of automotive charging circuits), to provide for a steady on of 4.5amps with interim 8 amp peaks, and to provide good isolation via an opto-isolator to a 5V or 3.3V uC. All this in a rugged design that should last in actual use. Certain Metal Oxide Varistors, like the 10N330K, can clamp spikes over 26VDC to further protect the high side, including TIP120. Another one should be considered to protect the the voltage regulator of the uC. So consider buying two of them.
Without a regulator, going through the 4N33, the current is restricted by a 130ohm, 2watt resistor being fed 10-15VDC. This a much simpler solution, but current to the TIP120 does drop when the supply voltage to motor drops. Nonetheless, it seem to work fine in a 10-15VDC range. One is always going to have varied voltages in an automotive charging environment. If that is considered a problem with keeping the TIP120 fully on, the 5 volt regulator that is independent of the uC is an okay option with a 47ohm 2watt resistor restricting its current.
I've pretty much done all that is required to be sure of a good design, but have also learned that I had several glaring misconceptions about the design transistor switching. So rather than continuing to provide bits and pieces in postings, I will write up a full explanation as an attachment (with supporting references) that justifies the role and size of each component, and then post it with schematics included for anyone that wants to use a 4N33 and TIP120 in such a context.
In that way, fellow learners can study the theory, while amateur builders can just copy the circuit.
Through out testing, I have use a series of fuses - 1.5amp, 2.5amp, 3.0amp, 4.0, and 5.0amp - to assure that what I was building wasn't going to dramatically go up in smoke or start a fire. I also started with a smaller load and later increased it. And still, about 6-7 of the fuses were blown. And a couple of voltage regulators and one 4n33 were destroyed by mistakes. Better doing this at a test bench with some control and learning why. I first tested circuits at 8 volts before going up to 14 volts for similar reasons. There is never a good reason for putting full voltage and full current to a device right away when you just want to be assured the wiring is correct.
If you feel the thread is too long and too complicated, just read the final write up and forget all the drama.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Ain't gadetry a wonderful thing?
aka G. Herzog [noparse][[/noparse] 黃鶴 ] in Taiwan
Post Edited (Loopy Byteloose) : 8/4/2010 9:52:27 AM GMT
h_FE (Beta) is what gets you where you want to go.· One should not have a beta-dependent situation and·ought to·have more base current than the ragged-edge provides, but not to an excess or, in your case, an excessive excess.··You're just burning stuff up.
All of this stuff·to date has been based on·misconceptions and worse; deciding where to begin is bewildering, but it's been instructive from a what-not-to-do perspective.
Unless this guy needs PWM then this whole trip is unnecessary.· A simple relay would switch his inductive load best.
IMHO a relay or a MOSFET . one can get Logic level fets with ease . I have not seen any for 3.3 V but with a small BJT one can fix that issue .
I allmost never use BJTs for switching anymore .. its IGBT or MOSFET or maby a relay ........
Peter KG6LSE
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
"Carpe Ducktum" "seize the tape!!" peterthethinker.com/tesla/Venom/Venom.html
Never underestimate the bandwidth of a station wagon full of tapes hurtling down the highway. —Tanenbaum, Andrew S.
LOL
Yes, relays are certainly a good valid alternative and there are some very small 30amp 12volt SPST automotive relays that could be driven directly from an 4N33 opto-isolator. These would certainly eliminate the heat factor if this is under a dashboard. But it would also add to the noise that might reach the uC in an automotive environment.
I happened upon a few things that made me want to pursue this.
The thread was inquiring about solid-state relays in DC. And, I recently built a rather wimpy TIP120 circuit off the web that I am now using to drive a bank of 4 12volt, 60ma relays. It was a TIP120 directly driven from a uC with a 470ohm resistor between. It works fine and runs cold, but I could have just as easily used a 2N2222a to do the job. I began to wonder why the TIP120 was recommended. Most of what I see on the inter-net under-uses the TIP120. And yet, we see many people try and fail to power huge devices with the UNL2803, which goes up in smoke driving less than an amp.
And then there was the idea to convert that existing board of mine to drive a rather hefty stepper of some sort. I wanted to see how the TIP120 handled heat and near full load. It is now obvious that the existing board will never handle power as it is too difficult to insert big heat sinks.
To help out, I explored the automotive context for a 'solid-state DC relay", but for myself I now feel confident about using the TIP120 in a large 12volt stepper or using the TIP120 with the TIP125 in a solid H-bridge that is micro-controller.
I have tried to be honest in order to present 'what not to do' and where blind alleys are.
For me personally, there were mistakes in my thinking and bad assumptions of what I was reading into the PDFs. Also, there was indeed one PDF with wrong information about the 4N25.
But I will still stick with what Horowitz insists about driving BJTs to saturation. He wrote a whole appendix on the topic in order to clarify. And I will sit down an down the Ebers-Moll Model calculation of what h_FE should be under this circumstances to see if I really got off into the weeds again. For that one needs to know that saturation current (Is) is at 5amps, the rest of the input is obvious. I just haven't had time fiddle with that. And it does indeed seem rather strange to have a Darlington that is reputed to provide h_FE of over 1000 operating at an h_FE of merely 7.
That is exactly why I keep investigating. Much of this thread seems like of blog about the muddle of depending on the web for good electronic info. That in itself may be useful to many.
P.J. I'd love to hear the right approach to sizing a similar 5amp 15volt transistor switch with the same parts.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Ain't gadetry a wonderful thing?
aka G. Herzog [noparse][[/noparse] 黃鶴 ] in Taiwan
Post Edited (Loopy Byteloose) : 8/4/2010 2:51:33 PM GMT
PJ,
I have no idea what the 'standard EE practice' is. I have a Bachelors in Arts in Fine Art. I've asked you to clarify a few times, but you seem to prefer not to. I would love to know the 'standard' way to do this. I have tried to interpret a well-accepted reference text. And I do suspect that I could somewhat reduce the TIP120 base current to make the device last longer.
As it stands the TIP120 has a gain of 41-42 (or between 6 and 7 for each of the paired transistor). Horowitz mentions switch with a gain of 10 to 20 is not uncommon - other bring the gain down to 5 for BJTs.
For 'standard EE practise", I just don't have:
1. the library
2. the education
3. the experience
And so I built a test bench to see what really works. Is that so bad?
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Ain't gadetry a wonderful thing?
aka G. Herzog [noparse][[/noparse] 黃鶴 ] in Taiwan
Post Edited (Loopy Byteloose) : 8/5/2010 7:20:03 AM GMT
I am not getting anywhere quickly with writing up an explanation for the design, but I can at least provide a final version of the schematic. The 10K resistor is NOT a pull down. It is there to prevent the base from being pulled below GND by a spike. (The TIP120 has an 8K internally, but why not be rugged?)
BTW, if heat is an issue, mechanical relays are likely better as both BJTs and MOVfets create quite a bit in a high power switching context. And I have mentioned there are some good cheap 20-30amp automotive relays that will likely work well with the 4N33. But becareful with specs. Originally, I had seen a PDF for the 4N33 that claimed 500ma, but now I am seeing 125ma output. Automotive relays can easily run up towards 200ma.
I have a series of more conventional relays I use that go up to 12 amps and require only 60ma. But it makes sense to use automotive relays in an automotive environment.
As always, it is all about finding a good fit. The LED on the opto-isolator is being driven by 10ma, but you can get by with less and suffer more sluggish response. Read the 4N33 PDF for details.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Ain't gadetry a wonderful thing?
aka G. Herzog [noparse][[/noparse] 黃鶴 ] in Taiwan
Post Edited (Loopy Byteloose) : 8/5/2010 12:38:25 PM GMT
Comments
The "1000 guaranteed" is the minimum h_FE specified in the datasheet which is·generally the last place anybody looks for advise or guidance on this Forum.· I see how these pesky PDFs·are... getting in the way.·
Wrong things are being "taught" here, but since you're "having fun" then I don't care.
I dug out my copy of 'The Art of Electronics' by Horowitz, et al. to see if I was getting off into the weeds. Searching via Google has not really been as helpful as I feel it should be and my library here is quite limited.
Apparently Horowitz is where I got inspired to drive the TIP120 at 100ma. I refer to pages 61-65 (particularly Section 2.02 - Transistor switch)and Appendix G, wherein he discusses misconceptions about h_FE and the benefits of driving transistors into saturation in the context of transistor switching (not amplification). He seems to explain that saturation with excessive base current is better for a transistor used in switching mode.
To quote him directly, "A circuit that depends on a particular value for h_FE is a bad circuit."
I will leave it to him to fully explain why, but transistors are not truly current amplifiers as represented by h_FE. He points out that they are 'transconductance amplifiers' wherein in a collector current is determined by the voltage applied to the Base-Emitter junction. He later goes into detail about factors that bend that linear relationship into a heavy curve by presenting the Ebers-Moll Model as his preferred means of determining gain -- variations in temperature, current output, and voltage output all change the h-FE substantially. And in a 'transistor switch' it is not unusual to see a simple transistor de-rated to an h_FE between 10 to 20 as great amounts of heat, higher voltages, and higher currents are applied. The darlington pair is not immune to such de-rating, but still provides a higher gain.
I do admit I may not have gotten my explanation of voltage drops correct and I am looking to that now. Doing this bench work has been very helpful to pointing out what I am not seeing in the PDF or what I am reading wrong. After all, if I read it wrong, something on the bench shows me that I should read it again or do more research elsewhere.
Incidentally, this is intended to run an inductive device a motor and that excess saturation may also be required to start the motor. The TIP120 tolerates 5 amps in a steady on, but up to 8 amps momentarily. So I am inclined to stick with 100-120ma driving the base.
The opto-isolator configuration seems a little odd to me, but I wasn't sure how to limit the TIP120's Base voltage to 5 volts. I must say that I do wonder if the voltage regulator is at all necessary. It might be much simpler to drive the TIP120 with a higher value resistor being fed by the automotive '12 - 15 volts'. Apparently this will work, and I will try it on the bench at the risk of burning up another 4N33. It certainly seems to better serve the concept of isolation.
Rather than not point out my mistakes 'because I am having fun', please show me what you feel are genuine misconceptions. Just because I am having fun, doesn't give me license to blindly convey misinformation.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Ain't gadetry a wonderful thing?
aka G. Herzog [noparse][[/noparse] 黃鶴 ] in Taiwan
Post Edited (Loopy Byteloose) : 8/3/2010 9:08:32 PM GMT
Well, "The Art of Electronics" was indeed helpful with pointing the way for me to design a powerful DC switch/solid-state relay equivalent. And yes, a power MOSfet is likely to be a good alternative to the power Darlington BJT. I may do some bench tests of a MOSfet at a later date. Still the TIP120 is very easy to buy and works well if you can tolerate the heat.
I have managed to eliminated the need for the 4N33 to be powered by a 5 volt regulator, so there is no need to run tests for a similar circuit with a 3.3volt regulator to provide nearly 120ma to the base of the TIP120. But a 3.3volt regulator should work with the right current limiting resistor value, say 28 ohms at 1 or 2- watt. (I am using 2 watt because the local store doesn't sell 1 watts, besides it gives me more useful low ohm resistors for other bench tests. Why buy 1/2 watt, 1 watt, and 2 watt values for testing, when a set of 2 watts will do it all?)
The design scheme has been intended to allow the 'high side" to go up to 15 VDC (which is safely beyond the normal limits of automotive charging circuits), to provide for a steady on of 4.5amps with interim 8 amp peaks, and to provide good isolation via an opto-isolator to a 5V or 3.3V uC. All this in a rugged design that should last in actual use. Certain Metal Oxide Varistors, like the 10N330K, can clamp spikes over 26VDC to further protect the high side, including TIP120. Another one should be considered to protect the the voltage regulator of the uC. So consider buying two of them.
Without a regulator, going through the 4N33, the current is restricted by a 130ohm, 2watt resistor being fed 10-15VDC. This a much simpler solution, but current to the TIP120 does drop when the supply voltage to motor drops. Nonetheless, it seem to work fine in a 10-15VDC range. One is always going to have varied voltages in an automotive charging environment. If that is considered a problem with keeping the TIP120 fully on, the 5 volt regulator that is independent of the uC is an okay option with a 47ohm 2watt resistor restricting its current.
I've pretty much done all that is required to be sure of a good design, but have also learned that I had several glaring misconceptions about the design transistor switching. So rather than continuing to provide bits and pieces in postings, I will write up a full explanation as an attachment (with supporting references) that justifies the role and size of each component, and then post it with schematics included for anyone that wants to use a 4N33 and TIP120 in such a context.
In that way, fellow learners can study the theory, while amateur builders can just copy the circuit.
Through out testing, I have use a series of fuses - 1.5amp, 2.5amp, 3.0amp, 4.0, and 5.0amp - to assure that what I was building wasn't going to dramatically go up in smoke or start a fire. I also started with a smaller load and later increased it. And still, about 6-7 of the fuses were blown. And a couple of voltage regulators and one 4n33 were destroyed by mistakes. Better doing this at a test bench with some control and learning why. I first tested circuits at 8 volts before going up to 14 volts for similar reasons. There is never a good reason for putting full voltage and full current to a device right away when you just want to be assured the wiring is correct.
If you feel the thread is too long and too complicated, just read the final write up and forget all the drama.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Ain't gadetry a wonderful thing?
aka G. Herzog [noparse][[/noparse] 黃鶴 ] in Taiwan
Post Edited (Loopy Byteloose) : 8/4/2010 9:52:27 AM GMT
All of this stuff·to date has been based on·misconceptions and worse; deciding where to begin is bewildering, but it's been instructive from a what-not-to-do perspective.
Unless this guy needs PWM then this whole trip is unnecessary.· A simple relay would switch his inductive load best.
I allmost never use BJTs for switching anymore .. its IGBT or MOSFET or maby a relay ........
Peter KG6LSE
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
"Carpe Ducktum" "seize the tape!!"
peterthethinker.com/tesla/Venom/Venom.html
Never underestimate the bandwidth of a station wagon full of tapes hurtling down the highway. —Tanenbaum, Andrew S.
LOL
I happened upon a few things that made me want to pursue this.
The thread was inquiring about solid-state relays in DC. And, I recently built a rather wimpy TIP120 circuit off the web that I am now using to drive a bank of 4 12volt, 60ma relays. It was a TIP120 directly driven from a uC with a 470ohm resistor between. It works fine and runs cold, but I could have just as easily used a 2N2222a to do the job. I began to wonder why the TIP120 was recommended. Most of what I see on the inter-net under-uses the TIP120. And yet, we see many people try and fail to power huge devices with the UNL2803, which goes up in smoke driving less than an amp.
And then there was the idea to convert that existing board of mine to drive a rather hefty stepper of some sort. I wanted to see how the TIP120 handled heat and near full load. It is now obvious that the existing board will never handle power as it is too difficult to insert big heat sinks.
To help out, I explored the automotive context for a 'solid-state DC relay", but for myself I now feel confident about using the TIP120 in a large 12volt stepper or using the TIP120 with the TIP125 in a solid H-bridge that is micro-controller.
I have tried to be honest in order to present 'what not to do' and where blind alleys are.
For me personally, there were mistakes in my thinking and bad assumptions of what I was reading into the PDFs. Also, there was indeed one PDF with wrong information about the 4N25.
But I will still stick with what Horowitz insists about driving BJTs to saturation. He wrote a whole appendix on the topic in order to clarify. And I will sit down an down the Ebers-Moll Model calculation of what h_FE should be under this circumstances to see if I really got off into the weeds again. For that one needs to know that saturation current (Is) is at 5amps, the rest of the input is obvious. I just haven't had time fiddle with that. And it does indeed seem rather strange to have a Darlington that is reputed to provide h_FE of over 1000 operating at an h_FE of merely 7.
That is exactly why I keep investigating. Much of this thread seems like of blog about the muddle of depending on the web for good electronic info. That in itself may be useful to many.
P.J. I'd love to hear the right approach to sizing a similar 5amp 15volt transistor switch with the same parts.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Ain't gadetry a wonderful thing?
aka G. Herzog [noparse][[/noparse] 黃鶴 ] in Taiwan
Post Edited (Loopy Byteloose) : 8/4/2010 2:51:33 PM GMT
I have no idea what the 'standard EE practice' is. I have a Bachelors in Arts in Fine Art. I've asked you to clarify a few times, but you seem to prefer not to. I would love to know the 'standard' way to do this. I have tried to interpret a well-accepted reference text. And I do suspect that I could somewhat reduce the TIP120 base current to make the device last longer.
As it stands the TIP120 has a gain of 41-42 (or between 6 and 7 for each of the paired transistor). Horowitz mentions switch with a gain of 10 to 20 is not uncommon - other bring the gain down to 5 for BJTs.
For 'standard EE practise", I just don't have:
1. the library
2. the education
3. the experience
And so I built a test bench to see what really works. Is that so bad?
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Ain't gadetry a wonderful thing?
aka G. Herzog [noparse][[/noparse] 黃鶴 ] in Taiwan
Post Edited (Loopy Byteloose) : 8/5/2010 7:20:03 AM GMT
BTW, if heat is an issue, mechanical relays are likely better as both BJTs and MOVfets create quite a bit in a high power switching context. And I have mentioned there are some good cheap 20-30amp automotive relays that will likely work well with the 4N33. But becareful with specs. Originally, I had seen a PDF for the 4N33 that claimed 500ma, but now I am seeing 125ma output. Automotive relays can easily run up towards 200ma.
I have a series of more conventional relays I use that go up to 12 amps and require only 60ma. But it makes sense to use automotive relays in an automotive environment.
As always, it is all about finding a good fit. The LED on the opto-isolator is being driven by 10ma, but you can get by with less and suffer more sluggish response. Read the 4N33 PDF for details.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Ain't gadetry a wonderful thing?
aka G. Herzog [noparse][[/noparse] 黃鶴 ] in Taiwan
Post Edited (Loopy Byteloose) : 8/5/2010 12:38:25 PM GMT