Opto-Isolator Oscillator

[@PhilC184]

Takes me back to my early days in electronics when oscillators would never oscillate but amplifiers invariably did

[@bigclivedotcom]

I think the optos are batch tested and ranked according to an average of their result. The current transfer ratio is the coupling gain from side to side, so for a CTR of 300% a current of 1mA on the LED side would result in a current of 3mA on the transistor side. Going by the spec sheet rank N is a bad day on the production line.

[@joinedupjon]

There's something a bit odd about those ranks... an opto with ctf 159% could go into 6 out of the 8 possible ranks.
I'd supposed they were aiming for about 150% CTF, tested samples of the batch for rank and sold whatever they had most left over of as rank N... if true you'd expect a lot more 300s than 600s in the rank K (300%-600%)
but they might be running different batches with process tweeks for each different rank... I've never worked in a semiconductor factory :(

[@gordonlawrence4749]

Certainly Toshiba used to test them as bigclive states and if I remember correctly from my days in distribution so did HP. Due to manufacturing techniques (more specifically the material used for the window and it's exact positioning) not only does average CTR vary but so does the distribution. EG if you have 100 on every percentile from 100 to 200 the average is 150 but if you got one on every third percentile from 0 to 300 it would still be 150 average (approximately) but they would be in a different category. The parts are not individually tested usually (the ones that are cost way more) and usually about 5% of a batch will be tested and they will be "binned" from there. EG bin H or K etc.

[@RWBHere]

Semiconductor device characteristics vary, depending upon a number of factors, including exactly where on a wafer the die was situated. We used to make specialist transistors for customers, and these were graded into bands, each being sold at a different price. They were usually selected from a known good processing batch, determined by initial pseudo-random sampling whilst still on the wafer.
When a known poorer batch was being processed, all of the finished chips were given a basic 'Pass/Fail' test, and then the working devices would all be sold under what the Opto-isolator manufacturer calls 'N' ('Non-selected', in our case). Within that batch, it is inevitable that some smaller percentage will meet the highest spec, and most will be around the bottom third of the spec range, with a few just exceeding the lowest specifications.
If the manufacturer has a glut of good wafer batches, more devices might be put into the 'N' category, simply because they have to sell the devices, so you might sometimes, as a customer, strike lucky. Because of the reduced amount of device testing, the price to the customer is lower.
I hope that this clarifies things for people?

[@gordonlawrence4749]

yes but hfe does not vary by anything like the amount that opto-isolators vary by.

[@Roy_Tellason]

I've also seen similar "grading" on some japanese transistor datasheets, where you would have a suffix letter that indictated a specific range of beta.

[@BEdmonson85]

Fantastic! I love to see unusual uses of different circuits and/or components. Here's a tip... If you need an non-polarized electrolytic cap, in a pinch you can use two polarized caps connected in series with either their + or - ends connected together. Just remember doing so will cut the capacitance in half. I have done this a number of times, in a pinch, with no problems. If you don't believe me, Google it.

[@GreenAppelPie]

This is electronics 101.

[@km5405]

- heres a tip: give the elilepsy warning before showing the flicker optoception oscillator of doom ™

[@Anvilshock]

Here's a tip: The thumbnail shows a row of lit LEDs, and the video title includes the word "oscillator". Now, I wonder what those two put together could possibly result in other than happy blinkenlights for you to still needing an explicit warning ...

[@km5405]

preferably fire and sparks : that is always exciting.

[@Anvilshock]

And fire and sparks would be less flashy?

[@start7047]

A bit late, but better now, than never ..
Frequency is "Everything"
"Everything" is Frequency

[@edgeeffect]

That is possibly the most aesthetically pleasing breadboard I've ever seen.

[@peteb2]

Was once sent to make repair on security camera pan&tilt in the field once. It was intermittent and the owners said music was coming from the 'camera box'. Once up the tower and with the P&T box open i actually heard music! The controller drove a relay card via optos that isolated it from the relays which inturn switched Vdc to the P&T motors and brakes. The card had previous repairs. Optos with far too much gain by the transistor had been substituted i guess and were acting as AM detectors, allowing just enough standby current through the relay coils to make them mechanically resonate. Unusual feature i told the owners...

[@nicnl255]

Is this the great return of the opto-isolator computer?

[@JulianIlett]

Maybe :)

[@adamthedog1]

Julian Ilett How about a relay computer? Noisy, slow, and fairly expensive; the perfect project.

[@Mark1024MAK]

adamthedog ​ Even better if it uses uniselectors :-)
czcams.com/video/BactVjevv_E/video.html
czcams.com/video/wWvImmZCuLQ/video.html
czcams.com/video/DXaJm0G4284/video.html
czcams.com/video/J3PpnUSxiSg/video.html

[@UpLateGeek]

If you wanted a larger capacitance, you could put two electrolytics of the same value in series with opposite polarities. So -++- or +--+. Only yields half the value of the individual cap, but would certainly allow you to go higher than 10uF.

[@chrismr3972]

A extra transistor on the output of the opto set to amplify current would do it. For the top opto it would be the opto emitter going to an NPN base and then the NPN collector to +5v. The extra transistor emitter would be the output. On the lower side the opto collector would go to a PNP base whose collector goes to 0v, again emitter to the output. Now you have lots of fan-out plus extra sensitivity!

[@trinidad17]

So it's an Opto-Oscilator

[@thewhitefalcon8539]

Or even an optiscollator.

[@Anvilshock]

Occliptostatistocitor?

[@James02876]

Because you're seeing a reduction in brightness on the red LED when you leave the final stage out of the ring oscillator, I would argue that the ratio is not above 200 percent, but is between 100 and 200. Because you see no reduction in luminosity when each stage is only driving one following stage, you've got at least 100 percent, which makes sense. However, if you had enough current to drive double the load on the output as compared to input (i.e. one input LED driving to output LEDs), you wouldn't see an indication of reduced current on the output. Since you can see it dim, you've got reduced current (assuming both output loads draw equal current in parallel). Since red LED forward voltage is at the lowest end of the range, it is a good indicator that the voltage has dropped due to increased current.

[@magnuswootton6181]

love the kickstart wire.

[@gregclare]

Hmmm. It seems a bit contradictory / self-defeating, to use an opto-isolator and then electrically bridge it with a capacitor! As you're no longer isolating the input from the output, you may as well save on LED power and just directly drive the base of the switching transistor. You could even come up with a descriptive name like Transistor-Transistor Logic, or even call it TTL for short. :)

[@JulianIlett]

The Vcc and Gnd bridges defeated the isolation long before I added the capacitor :)

[@BertGrink]

Isoscillator
Re. the capacitors: there are bipolar electrolytic caps, made especially for audio circuits, but they tend to be a bit pricey; however it should also be possible to hook up two regular elco's in antiparallel and have the same effect

[@qwertyasdf66]

I can't be the only person who thought "Oh, sweet, a redstone circuit!".

[@charliedobbie8916]

Torch'll burn out if it flashes this fast...

[@matteopascoli]

And a pressure plate for the kickstart ;)

[@Gooberslot]

The world's blinkiest computer... This needs to be done.

[@charliedobbie8916]

You had me at "blinkiest"

[@Sparkette]

I never noticed until now how similar those two words are.

[@Derundurel]

You can use two electrolytic capacitors "back-to-back" for larger values, but it isn't so elegant. You should also add a bias resistor from the common negative terminals to the negative rail. Alternatively, you can join the positive terminals and use a pull-up resistor. The value isn't critical, but 1 MΩ should be fine. I have seen this done without the resistor, but then you have an unpredictable bias across each capacitor and consequently greater uncertainty in the overall value. Reliability will also be lower. One disadvantage with this technique is that the overall value is only half the value of the two capacitors.

[@krishna34674]

FINISH THE VOCODER ALREADY !

[@JulianIlett]

He he, how about a version using op-amps made of discrete components :)

[@krishna34674]

AAAAAAAAAA ! How about a version using all SMD parts and inexpensive custom PCBs from china ?

[@nicholasroos3627]

so much yes. so much

[@PrinceWesterburg]

Valves!

[@God-CDXX]

I am trying to do a VOCODER with TUBES

[@xanataph]

You can purchase bipolar electrolytic caps for larger non-polarised capacity. And as someone earlier commented, simply two normal electros back to back will do the trick too. I suspect that's how the bipolar ones are more or less constructed internally.

[@charliedobbie8916]

Some non-polarized electrolytics are *literally* two polarized electrolytics, with individual markings and everything, stuffed inside a larger can.

[@xanataph]

That doesn't surprise me! You would actually be able to take it apart and use each capacitor separately. Although you would have to be pretty desperate to do so.

[@MitkoNikov]

Interesting circuit as always and I love the neatness on the breadboard.. Although I would prefer working on the Vocoder, it's such a nice project sitting on some shelf... Keep it up!

[@holtzda]

With the capacitor connected from the input to the output of DCOI stage, the effective input capacitance of a stage is the capacitance you added multiplied by the gain of the DCOI stage during the time when it is operating linearly. This is due to the Miller effect: Cin=C(1+gain). It's not just the doubling of the charging voltage that you noted that is important, but also the gain of the stage. Could you please measure the gain of a stage so that we can figure out the Miller input capacitance? Of course, this gets more complicated because your DCOI stages aren't always operating linearly.

[@Roy_Tellason]

Build this with IR LEDs instead of the red and green, power it with a couple of coin cells, stuff it into a hatband with the LEDs sticking out, and watch what fun things it does to surveillance cameras... :-)

[@SomeMorganSomewhere]

The "ranks" will be how the ones they actually bother to characterise are binned, the "N" rank will be "everything else"

[@T2D.SteveArcs]

Thats really good Julian, You could have used two electrolytic caps back to back to form larger non polarised caps ....

[@coxsj]

Fun vid, thanks for sharing.
Your "Pound Shop" comment confused me for a moment till I realized that is the UK version of the 'Dollar Store'! Separated by a common language! Cheers!

[@jstro-hobbytech]

you could make a handful of jellybeans oscillate. what an awesome circuit

[@ToddFun]

You should put the out put through a audio amp to a speaker. Then pull out the caps so we can hear the freq changes. That'b be fun!

[@KX36]

Here's a "trap for young players". The charts in datasheets for optoisolators are often for *Normalised* CTR. Just look out for that word because it's so easy to get all your calculations way off, especially as CTR and Normalised CTR are often on the same order of magnitude. Other thing about optos is that the bandwidth is proportional to the current.

[@monad_tcp]

so I should drive them at 3A, sure that would get me a big bandwidth. (I'm kidding, the bandwidth will be 0)

[@Pops180]

You can use 2 electrolytic caps in series but with their negatives tied together

[@GordieGii]

If you make a version where the input is biased, it should self start. It only needs to be the first stage.

[@raymondheath7668]

Never once thought of using opto's for a ring oscilator, what made you think of it?

[@JulianIlett]

When I discovered (from the NEC2501 data sheet) that the CTR could be well over 100% :)

[@michaelc9009]

Those sure look like 10uf dipped tantalum capacitors. If so, they are extremely sensitive to polarization. Cool video!

[@drteeth7054]

The breadboard layout is not symmetrical at all. It just has repeating units.

[@gordonlawrence4749]

Technically it's a totem-pole arrangement as per the old days of TTL logic output stages. Complimentary would usually mean NPN+PNP.

[@JulianIlett]

TPCOI - hmm :)

[@-yeme-]

Im kind of amazed that you can get 10uF ceramics that small. I expected them to be 105s
you could parallel them up to get a higher value

[@power-max]

You can get capacitances on the order of 50uF to 100uF in that approximate size surface mount. I got some that are 47uF 16V in a 1210 SMD package.
The catch with the really high capacitance units is the dependence on voltage (tends to drop sharply with DC bias (usually class II and class III dielectrics like X5R and Y5V), also some piezoelectric effects.

[@DAVIDGREGORYKERR]

Wow the basis of a Dan Tayloe Mixer for use in the VOCODER.

[@SidneyCritic]

If they are Ebay - Ali is worse - parts they might not be genuine, i.e. a cheaper part in another case. They will work, it's just that it might not actually have the same specs as a genuine NEC opto.

[@zylascope]

Very Cool,Thanks Julian :)

[@mikehensley78]

I wonder....just using 1 pair of opto-isos instead of a cap in the middle could a motor be put there and used as an H-bridge? I know it wouldn't be very strong as far as amperage.

[@johnstephenson2891]

So a DCOI computer would be larger than the first one I worked on! Elliott 803b!! lol

[@boldford]

One could be forgiven for considering you to be OCD following your waxing lyrical about the symmetry on the breadboard .

[@gcewing]

Optoisilator Compulsion Disorder?

[@foxabilo]

Been having a lot of good results from HCPL-4701 Broadcom opto's , they do a whooping 3500% transfer and will do 400% with an input as low as 40µA .
You could even be cheeky and call it an optical computer. **whistle**

[@NNNILabs]

Now there's something cool!

[@meisam9592]

This sound much like one of the “Rhett and Link” “will it …” episodes!

[@paparoysworkshop]

What about using Tantalum Electrolytic Capacitors. They would tolerate the reverse current to a degree. Not in a finished product, but for the purpose of clarification in this video, I think they would survive long enough. Just a thought.

[@gorak9000]

Add more ceramic caps in parallel at each stage!

[@rbmwiv]

Very nice looking

[@californiakayaker]

Feeling very Opto Isolated here....

[@TauCu]

The Opto-Oscillator

[@peterjones2411]

interesting concept ! and you could use standard electrolytic capacitors connected in inverse parallel eg negative to negative to create a non polar electrolytic capacitor. you where I'm sure already aware. and to save a very long winded comment it's best looking up the explanation. it's a very old trick.

[@peterjones2411]

that should of said inverse series not parallel. sorry

[@deathshaker0026]

Thats too cool!

[@markflack9191]

Make a longer ring and add more capacitors to slow it down more.

[@raykent3211]

Sorry for being a grumpy git but I worry a bit about how beginners may be misled. A ring oscillator needs inverters and you can get 6 of them in a 4069, or similar package. That replaces 12 of your optoisolators and with much lower current consumption and cost. The only reason for using an optoisolator as an inverter is to get galvanic isolation between the input and output sides, which this circuit neither needs nor provides. I enjoy fanciful imaginative stuff, but at some point you should say that this is simply worse in every way than the standard approach. Otherwise seasoned engineers may chuckle in their beards while young uns get the wrong idea.

[@joinedupjon]

Yes you are being a bit grumpy imo - it's a playful, fun and quirky channel that revels in 'thinking outside the box' and not necessarily the place for learning about doing things correctly.
couple of weeks ago I did a search on 'play' and 'fun' on the channel and julian uses these words in his video titles quite often... maybe not often enough - perhaps people would be get less triggered if he used 'playing with...' 'fun' etc. in titles to divide the tutorial type video from the more lighthearted ones?

[@ufohunter3688]

Is this the clock for your new "discrete-built" computer? OR just a curiosity?

[@theoriginalbabycub]

Hi Julian. Your breadboard jumper wires (with the pins) look much better quality than any others I've seen. Can you tell me where you got them? TIA.

[@JulianIlett]

I cheated - these are old Maplin ones :)

[@StevenIngram]

Hmm, since this is bidirectional - could you invert logic by inverting polarity? I mean, if you can - it seems like you could reduce the instruction set of a DCOI computer right off the bat. And if inverting polarity could do inverse logic... could a well placed flip flop do converse logic? Ok, I admit it... maybe I don't understand what I'm seeing. LOL

[@NetworkXIII]

Julian, how about a computer built with nothing but DCOIs and 555s?

[@oldtimeengineer26]

Nice video

[@fute27]

I would be curious to see where zybee fits into your graphic... (3g/wifi/lora)...

[@joinedupjon]

9:06 You'd think the ranks would be priced differently but having a look at mouser they seem to be priced the same... don't suppose julian pays mouser prices for his optos though.

[@JulianIlett]

+joinedupjon 50pcs for 99c as I remember it.

[@joinedupjon]

if it says KK 433 on the second line I think they're the rank 'K' (300-600% CTR) components made in Taiwan - if we believe the markings.
fwiw mouser says that due to a takeover by Renesas the NEC branding has been replaced by a capital letter R

[@johncherry108]

During manufacture of the opto-isolator chips, they would be tested and various parameters measured so they can be 'ranked'. They would all take the same time to test and therefore the price would be the same.

[@jaro6985]

Sure they take the same time to test, but the demand for lower CTR would be lower so you'd think price should be lower as well. Similar to binned CPUs.

[@BertGrink]

As i watched the video, i also thought of the comparison to binned CPUs, however since these optoisolators are way cheaper than CPUs, the price difference between different ranks should be comparatively smaller too, reulting in roughly, or perhaps precisely, the same price for all ranks. Just a thought though.

[@mikerhodes9198]

Looking at the video at the first when it started oscillating, I see an inconsistency in the cycling LEDs. I am not sure if this is a conflict between the camera and frequency of the flashing LEDs or not. It appears to hang momentarily. Do you see this live or is it just a camera anomaly?

[@JulianIlett]

Probably an editing anomoly.

[@realflow100]

make an opto-isolator calculator with shift registers and logic gates or something with them!!!

[@gcewing]

So it's really a Dual Complementary Tristate Opto-Isolator... hmmm, probably best not to try to pronounce that.
On the plus side, with DCOIL you would be able to build non-inverting gates just as easily as inverting ones. There's no reason the upper opto-isolator needs to switch the positive side and the lower one the negative side. Swap the transistor connections and you have a non-inverting buffer.
Maybe the fanout problem could be mitigated by connecting opto-isolator inputs in series. If you eliminated the LEDs you should have enough voltage to drive a few inputs that way, as long as you have at least 100% CTR. Of course it wouldn't be so blinky that way. :-(

[@phonotical]

... aren't you lord buckethead...

[@JulianIlett]

aren't you

[@taranagnew436]

what projects r they used in, useage?

[@Derundurel]

It'd be interesting to see a processor built using this logic. The simplest I know of is the Motorola MC14500 series. It's a 1-bit design.
Datasheet: www.brouhaha.com/~eric/retrocomputing/motorola/mc14500b/mc14500brev3.pdf
Instruction set: en.wikichip.org/wiki/motorola/mc14500/isa

[@NetworkXIII]

Merlin Skinner These are selling for $2.50 or so on eBay, another possible spare-time project for the list.

[@RottnRobbie]

Buy rank 'K' parts from NEC, and and you'll get a long life out of that DCOI computer.
Your minimum CTR will be 300, so after degradation to 2/3 of that - i.e. to 0.666..repeating - it'll still be 200% or better. From the long term degradation chart, at 25°C it looks like that point will be reached after about 10^5 hours (a bit over 114 years) of operation!
Not sure about the life span of those caps, but you'd probably not want to artificially slow down the computer anyway, except to make the blinking visible to us poor humans. ;)

[@schimpl94]

I am very interressted of seeing this built on a PCB ! :)
There you do not have that high parasitic capacitance.
And what do you plan to take as output buffer ?
The SlewRate doea not look that great ...

[@Dellenite]

Would that be an ideal set up for parallel execution?

[@ThatGuy-nv2wo]

Did you consider adding a resistor to that cap to get a lower frequency? It would probably mess up how nicely layed out it is though

[@peterleather6866]

That idea of a resistor in series with the cap jumped out at me too. It seems the logical thing to try. I don't know if it would work though. Any thoughts Julian?

[@ThatGuy-nv2wo]

On further thoughts, I don't think it would if a resistor was simply places there, the input would instead have to be moved after the resistor but before the capacitor but that could have some effects with negative voltages?
No wait just use 2 resistors...I think

[@peterleather6866]

Surely the resistor would just slow the charging and discharging of the cap in either direction? Oh I don't know...my brain hurts now! :-)

[@ThatGuy-nv2wo]

Yeah it would wouldn't it?
I think I was just overthinking this haha, sorry!

[@tctrainconstruct2592]

3:45
Use caps in parallel!

[@flymario8046]

well that is pretty.

[@kluzz]

It's repetitive, not symmetrical. Still managed to enjoy the video, though!

[@mfx1]

Non polarised electrolytics exist.

[@WilliamDeboer]

ive been in a couple spots where i have to cover one eye and look away. o.O

[@maicod]

So the opto is 'spent' after 1000 hours ?

[@JulianIlett]

Not if your CTR requirement is less challenging :)

[@maicod]

you mean that if it resonates (on/off) at least 20 hz (20 times a second for 1000 hours) then they are 'spent' huh :) BTW I'm wondering is it the optocoupler's DIE burning up or getting insensitive or is its IR-LED 'dying'

[@DaleBasye]

what's the IC #

[@DaleBasye]

kewl PC817 ordered 50......

[@bobyla13]

lot off bla bla..so skip