Speaking of Retro
Humanoido
Posts: 5,770
How many vacuum tubes would it take to
build the equivalent of one Propeller chip?
I envision a kind of retro science fair project
for pure hobby and fun. Seems a good
question for all those vintage ham radio
old timers that are "hybrid" with the Prop.
humanoido
build the equivalent of one Propeller chip?
I envision a kind of retro science fair project
for pure hobby and fun. Seems a good
question for all those vintage ham radio
old timers that are "hybrid" with the Prop.
humanoido
Comments
-Phil
Leon
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Amateur radio callsign: G1HSM
Ha, Ha, I love the good old days.
I have a tube radio docking station for my Ipod and I love it, playing with tubes are fun !
Here is a pic of a similar type I have. I could not find the model I have on there site. It was given to me as a Christmas present in 08.
Here is the link: http://www.acoustic-research.com/homedecor/HomeDecorProductDetail.do?ACTION_TYPE_ID=ACTION_TYPE_PRODUCT_DETAIL&ACTION_CATEGORY_ID=STR_SEARCH_RESULT&search=tube &ACTION_PRODUCT_ID=AV100C
Post Edited (Rob7) : 2/17/2010 7:01:12 PM GMT
I remember my old Stewart Warner tube radio, Knight Kit Star Roamer, and tube built ham radio gear, and I think those tubes lasted a relatively long time. It would be no problem to stock some spares and replace any burned out ones. It's just like running a light bulb. It's not going to last forever.
Heat removal? Oh, I would run it only in winter so that can be a good thing. Position it in a skyscraper in the dead of winter and the bonus spinoff technology is free heat for all 75 floors! That's going to make everyone happy when their heat bill never arrives! [noparse]:)[/noparse]
humanoido
-Phil
I always wonder why there isn't a law that states that the heat wasted from the power station cooling stacks has to be piped around to the local comunities. Just up the road from here is an eight pot cloud generator.
Some of the transmitter stations I used to work at had meters on the walls showing the MWs incoming (and most of the outgoing was just propaganda!)
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Style and grace : Nil point
Anyway, wasn't the electro-mechanical relay used before the vacuum tube? I have a 4 bit adder circuit somewhere, trouble is it's about a foot square, I think a single 32 bit ALU would take most of a house let alone 8 of them.
SRAM: 32KB + 8*2KB = 48 * 8 Kbits = 393,216 bits. A ram cell takes 6 transistors = 2.4M transistors
ROM: 32KB = 32* 8 Kbits = 262,144 bits. A rom takes 1 or 2 transistors?? = 0.26M or 0.52M transistors
CPU: 8 * 32 bit processor and counters. My best guess (no idea really) = 1M transistors
IO, etc: Estimate = 0.1M transistors
So I am way off on my original guess as this is around 4 million transistors.
Now, a valve (vacuum tube) is 1 transistor, so we have 4 million valves.
The minimum space for a simple valve would be 1"sq disregarding heat dissipation.
So, we now have 4,000,000"sq of space without wiring and power supplies and access walkways. This is about 28,000 sq ft.
Assuming we house them in racks 8ft high (8ft usable plus rack height), we have 3,500 lineal feet of racks which is over 0.6 mile long.
Now if we are really ambitious we would have them in 50ft long bays... so 70 bays x 50ft
We will require mobile electric lifts with an OH&S chair for the operator to change his 50x8 foot bay of 57,000 tubes (forget the valve computer to run this LOL)
Based on Phil's calculations, we will be replacing 200 * 4 = 800 tubes an hour. Presuming we have a consistent dispersement of failures, each bay will experience 800 / 70 = 11.4 tube replacements per hour or more precisely, 1 every 5 minutes.
Now, if we run the cpu for 8 hours a day, we require 800 * 8 = 6,400 valves per day. Now where are we going to store the replacements and how are we going to handle shipping in and out 6,400 valves per day. If we run 24 hours per day, then almost 20,000 valves per day !!!
Next, we have 800 failures an hour = 13.3 per minute = 1 failure every 4.5 seconds.
And we haven't even discussed time to diagnose which valve is faulty. And while we are doing that, we are having compounding failures disrupting our diagnosis.
Houston, we have a problem
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Links to other interesting threads:
· Home of the MultiBladeProps: TriBlade,·RamBlade,·SixBlade, website
· Single Board Computer:·3 Propeller ICs·and a·TriBladeProp board (ZiCog Z80 Emulator)
· Prop Tools under Development or Completed (Index)
· Emulators: CPUs Z80 etc; Micros Altair etc;· Terminals·VT100 etc; (Index) ZiCog (Z80) , MoCog (6809)·
· Prop OS: SphinxOS·, PropDos , PropCmd··· Search the Propeller forums·(uses advanced Google search)
My cruising website is: ·www.bluemagic.biz·· MultiBlade Props: www.cluso.bluemagic.biz
I went to undergraduate school where they used Univac computers (www.fourmilab.ch/documents/univac/case1107.html). They also had a Univac I that had better up-time than the transistorized Univac 1107 shown in the link. Partly that was due to redundancy. I think there were essentially 3 arithmetic units that voted on their output.
There's no way you'd even try to do the memory with active devices. One you'd implemented the memories with core or disk or drum memory, that'd cut down the number of valves needed substantially (and the speed too). Doing 8 x 32-bit arithmetic units with barrel-shifters would be tough enough.
In 1989 I could still get them new at Radio Shack for $15, now they would be over $50 each.
If you live near a neon or glass art shop, it may be worth it to make your own.
But nobody wants to make a computer out of them.
Vacuum Fluorescent Displays are so tube-like that they may have alternate uses as tube integrated circuits.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
For me, the past is not over yet.
We use rotating magnetic drum for the HUB memory. With modern engineering it must be possible to get 8192 times 32 bit words on there.
We us magnetic core store for the COGs registers. They are not so big.
That only leaves the compute parts in tubes.
Now here is the magic part. The drum memory has 8 sets of read/write heads equally spaced around it. One for each COG. Bingo! We have a "mechanical hub". No extra logic required.
So how many tubes are we down to now? Given the simple and regular nature of the COG instructions this looks quite doable.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
For me, the past is not over yet.
Well the·magnetic drums·that our overseas·Telecom (OTC in those days) used in the early 70's for storing and forwarding telexes were about 4ft x 4ft x 10ft. They had two and two Univac computers. One was always hovering after the other waiting for a crash so it could take over. Then came the crash dump via the Univac 1004 printer. Then it was a manual bootstrap entry by switches to get a mini program entered that would boot the next level from paper tape which then loaded the real program from punched cards·via the Univac 1004. IIRC booting took about 30 minutes and you hoped the other computer stayed up longer than that
I think both core memory and magnetic drums are cheating. The Magnetic Drum could be substituted for our microSD card and the core memory could be substituted for our external SRAM. But hey, it's retro days, and they did not exist!!!
Note to self: Get your head out of the clouds and back to real prop stuff
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Links to other interesting threads:
· Home of the MultiBladeProps: TriBlade,·RamBlade,·SixBlade, website
· Single Board Computer:·3 Propeller ICs·and a·TriBladeProp board (ZiCog Z80 Emulator)
· Prop Tools under Development or Completed (Index)
· Emulators: CPUs Z80 etc; Micros Altair etc;· Terminals·VT100 etc; (Index) ZiCog (Z80) , MoCog (6809)·
· Prop OS: SphinxOS·, PropDos , PropCmd··· Search the Propeller forums·(uses advanced Google search)
My cruising website is: ·www.bluemagic.biz·· MultiBlade Props: www.cluso.bluemagic.biz
I was in the Ontario area last week and saw a new 100,000 ft sq building for rent.
With an unlimited budget, when do we start !
My last valve project was rebuilding my 2m FM transmitter in 1970. I used a FET front-end receiver and as my electronics teacher used to say, a fet is only a valve without the pilot light LOL.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Links to other interesting threads:
· Home of the MultiBladeProps: TriBlade,·RamBlade,·SixBlade, website
· Single Board Computer:·3 Propeller ICs·and a·TriBladeProp board (ZiCog Z80 Emulator)
· Prop Tools under Development or Completed (Index)
· Emulators: CPUs Z80 etc; Micros Altair etc;· Terminals·VT100 etc; (Index) ZiCog (Z80) , MoCog (6809)·
· Prop OS: SphinxOS·, PropDos , PropCmd··· Search the Propeller forums·(uses advanced Google search)
My cruising website is: ·www.bluemagic.biz·· MultiBlade Props: www.cluso.bluemagic.biz
over 8 hours a day, the ham radio gear left on most of the day,
the various SWL radios on about 2 to 4 hours a day, the amps
were on about an hour to two hours a day, and overall
there was not one single tube replaced in ten years!
If each new tube will last 5,000 hours, then it does not matter
how many tubes you have, they will all last 5,000 hours from the
time they are switched on (defects aside). For example,
use our retro computer for 5,000 hours before replacing tubes,
and fire it up one hour per day, say 3 times a week, with 52
weeks in a year, thats 156 times of use per year which equals
a mere 156 hours, which will give us 32 years of use.
For a hobby retro computer project using tubes, IMO, 32 years
of tube operation time is more than I personally need.
Even if you cut this figure by five times, it will still run for six years
before you need to replace tubes.
humanoido
So thats everything sorted out on the space issues, but it's just doubled the chances of a failure.
Yeah, bored at work. Not 51 anymore
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Style and grace : Nil point
Sadly, that is not how things work. Ever hear of the "bath tub" curve? en.wikipedia.org/wiki/Bathtub_curve
Think of tube failures more like radio active decay, you can never tell when they will go off.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
For me, the past is not over yet.
It was discussion about a theoretical " defects aside" at first for the 5,000 time frame.
Next, considering these tubes are like light bulbs and not exactly predictable
when they will go off, you can see it was divided - 5,000 hours by five to
compensate for these anomalies. Yes, absolutely we can never predict when one
will go off in the real world. One suggestion is going with military spec
which pre-stresses the part to assure greater predicted longevity.
I would also suggest, after the tubes are purchased, they could be put through
a jig that can do stress induction, to weed out the failures ahead of time.
en.wikipedia.org/wiki/Defense_Standard
humanoido
Post Edited (humanoido) : 2/18/2010 10:12:36 AM GMT
I fear a buget overspend ...
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Style and grace : Nil point
no problem, we'll brew up a batch of our own tubes...
remember that glass blowing kit we got as a kid?
We can put that knowledge to good use..
humanoido
When the vacuum was required we could put the valve up to our lips whilst somebody told us the price that Farnell and RS want for the Props here in the UK. With such a sharp intake of breath we would be picking bits of gettering out of our teeth.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Style and grace : Nil point
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Visit some of my articles at Propeller Wiki:
MATH on the propeller propeller.wikispaces.com/MATH
pPropQL: propeller.wikispaces.com/pPropQL
pPropQL020: propeller.wikispaces.com/pPropQL020
OMU for the pPropQL/020 propeller.wikispaces.com/OMU
pPropellerSim - A propeller simulator for ASM development sourceforge.net/projects/ppropellersim
If you think you want to handcraft your own tubes, be sure to watch this video first:
····www.dailymotion.com/video/x3wrzo_fabrication-dune-lampe-triode_tech.
Making a million of them could take awhile.
-Phil
That is one of my all time favorite videos. That is dedication - its nuts - but its dedication.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Whit+
"We keep moving forward, opening new doors, and doing new things, because we're curious and curiosity keeps leading us down new paths." - Walt Disney
humanoido
Graham