Designing a classic transistor-VCA from scratch
Вставка
- Опубліковано 23 лип 2024
- Support the channel...
... through Patreon: / moritzklein
... by buying my DIY kits: www.ericasynths.lv/shop/diy-k...
Working simulation on falstad: tinyurl.com/ygbk9l6f
In this double episode, I’ll walk you through the process of designing a classic transistor-based VCA (voltage controlled amplifier). The goal is to first de-mystify the bipolar junction transistor (BJT), its usage in a common emitter amplifier plus its two central flaws (variations between seemingly identical samples & temperature dependence), before then exploring the differential amplifier (also known as the long-tailed pair) and how we can get it under voltage control.
Inverting amp in-depth explanation: • DIY SYNTH Quick Tip: S...
AC coupling in-depth explanation: • Adding voltage control...
If you want to follow along, here’s a list of components:
2x BC548 NPN transistor (matched)
1x TL072 dual op amp (TL074 works too)
1x 100k potentiometer
2x 470k resistor
1x 200k resistor
3x 100k resistor
1x 33k resistor
3x 10k resistor
1x 1k resistor
1x 100 ohms resistor
Chapters:
00:00 Intro & Sound Demo
03:53 Voltage Dividers
09:22 Resistors vs. Transistors
14:12 Common Emitter Amplifier
19:50 Emitter Resistors & Negative Feedback
27:34 Gain Changing & Sketchy VCA
34:20 Diffamp/Long-Tailed Pair
39:11 Voltage Subtraction
43:41 Final Circuit
45:30 Sound Demo & Outro - Наука та технологія
Love the analog deep dive. So refreshing to get away from the usual "buy these modules and connect them like this" type of electronics tutorial.
To think we used to say the same thing about ICs and tubes lol
can we just stop for a moment to appreciate how great that intro jam was?
That intro jam came out of nowhere and was epic
This is UA-cam and the Internet at its finest. Thanks for the time you invested into this, and sharing your knowledge in such an accessible way and without losing any of the technicality. Subscribed.
Been messing around with diy stuff and reading about electronics for some time now but this channel is the first time I'm actually starting to grasp how all this really works. Awesome content
I really hope you keep making videos for a long time. The information you share is invaluable to all your followers. It’s obvious you go to great lengths to deliver helpful lessons so thank you sincerely.
Seeing the thought process of designing of such an analog circuit is inspiring. Thank you!
electrical engineer & musician here. that was the best explanation of analog circuits i've ever heard.
Great video, thanks for the in-depth explanation - I wish there where videos like this when I struggled with the basics of electronic circuits back in the days. This is hands-down the best description of the differential amplifier circuit I have come across.
Thank you for sharing this concept Moritz! You manage with ease to explain the functionalities, so you're not only inspiring to the synth diy community but also to anybody that wants to understand electronics!
This is awesome! I got done with building my own Minimoog clone last year and I had reverse engineered nearly every part of those schematics to understand their function in detail. The only part I didnt fully reverse engineer was the VCA. I mainly didnt understand the reasoning of why it needed to be a diff pair but now I understand that its to improve gain and get rid of the DC offset! Great video!
New subcriber of your from the philippines now i understand the working of vca for the amplifier to work through their bias voltage i have no training in electronics school thru presenration i understand how it works the circuit more power to your channel tnx sir
For a brief moment in time I thought that you were going to compensate the amplification error with am inverted feedback and then you came suddenly with something that is very clever and of which I am not able to see the delicate finesses. I clicked 27 times on the thumbs up, you deserve it.
Thanks Moritz for the another great video and explanations. Getting back on the breadboard and build this. have a nice day.
Moritz, I am a huge fan of your videos. I am always looking for example systems that are conceptually understandable to show to my students. Your explanations are always clear and intuitive -- I'm sure you put a huge amount of effort into making them so.
That said, the pressure/voltage analogy breaks down in this application. In the diagrams at 4:00, it is (in general) not correct that the three pipes would have different flow rates. The pressure halfway down the 2-constriction pipe would not be 1/2 of the pressure at the top. The breakdown of the analogy is due to the fundamental difference between the two quantities: voltage is proportional to a potential energy, whereas pressure is related to forces applied by/on the fluid. It's fine to think of both as "causing" flow -- pressure actually pushes the fluid from one location to another, [positive] charges move from high V to low V -- but pressure doesn't get "used up" in the same way that voltage does. Pressure is probably more analogous to the electric field in the wires/components, but this is (probably) wayyyy too low-level to care about when building an awesome device.
Pressure and fluid flow can be very counter-intuitive. I had to stare at the diagram for a good 5 minutes before I realized what wasn't quite clicking.
absolutely fantastic .perfect for someone like me who has been addicted to modular for a bit over a year, but knows dog shit about the technicalities AT ALL! das sind wunderbare Videos Moritz. 1000 Dank!!!
Love these videos and the depth of explanation. I know nothing about electronics but its very interesting
Love these synth module builds!!!
love you profile pic :D best album ever!
Once again pure gold Moritz. Good job sir, good job!
Hi Evan,
If you're interested in Hardware and interesting questions asked in the industry, please check us out! ua-cam.com/channels/7h3PROcX7Zgx00alQokJ-w.html
very excited to watch and try building this when uni isn't kicking my ass, your work is inspiring man
^v^
haha so true
Wonderful and simplified explanation
Thank you✨👌
Dude, I love your content! Really inspires me!
I like how he’s always using op amps to do almost everything 😎 those things can dam near do anything 😯
Forreal, once you understand how opamps work (actually once you start seeing them as simple comparators) you can design so many circuits. Opamps are awesome and devilishly simple
what affordable method of supplying the + and - voltages needed for op amps would you recommend? i've spent thirty years looking at op amps and thinking they are bastards because they make circuits so complicated to supply them and thus build.
i should probably stick to dsp, where things make sense :)
@@atomictraveller Depending on how you bias them, you can just sit them between ground and VCC (capacitor couple the output to remove DC offset). Otherwise, power supplies with + and - rails are fairly easy to find, and you could even just use two normal isolated supplies in series and ground the middle to get two rails
@@atomictraveller Depends on your application, but I usually bias the opamps with a zener diode. I found that the easiest way to get a stable biasing voltage. If you don't intend to saturate your opamps (like you shouldn't really) it doesn't even really matter if the zener isn't bang in the middle.
EDIT: when I said 'depends on your application' I meant if you're using it in eurorack and/or want to amplify DC voltages. For 99% of cases in audio you don't need to have a split rail in your opamps. If you really need a split rail it gets a lot trickier, but you can use complimentary LM78xx and LM79xx for a split rail if you really must. For breadboarding you'd want to invest in a (used) lab supply with split rail.
@@atomictraveller you can try to use icl7660
Mr Klien, please keep going, on and on, fantastic work, many thanks
Hi Cooked,
If you're interested in Hardware and interesting questions asked in the industry, please check us out! ua-cam.com/channels/7h3PROcX7Zgx00alQokJ-w.html
Fantastic video, thanks so much. Really enjoy your work.
Bob Moog would have been proud of you for sure.Thank you for this channel.
Looks great! I have built your ADSR on my modular setup, this one will join it. You can't really have enough VCA.
Absolutely incredible video, thank you!
maan i really like your taste. Video style, music, electronic, everything you touch is gold.
btw you could cover S&H module in the future video
will do!
I love the graph paper explanations.
Really solid theoretical explanation
We did it differently in the 1980s. Same schematic, but audio was always fed into the tail CCS, and CV into the differential pair. Otherwise, as shown in this video, it's an unfortunate tradeoff between very high distortion and very high noise (100k and 470k resistors in this example). Of course, feeding CV into the differential pair means that the control law will be influenced by temperature, but this is easily reversed by tweaking the CV scale. The double-photoresistor optocoupler is probably the best for audio. These were quite costly forty years ago, but today they cost pennies (although finding a closely matched pair is still difficult).
I'm also curious why the 100k/100R divider at the base of the input as drawn. Dropping the input down by a factor of 1000 only to bump it back up again with a [noisy] op amp.
I realize this is in the spirit of a synth and that distortions are less of a concern, but this isn't the only use case for a VCA, and where clean is important, this would fail to deliver as drawn.
I'd bump the 100R to a 1k and then just give the signal a little nudge with the op amp to bring it up to spec. Much cleaner.
the explanation is really nice
thank you for sharing
Amazing, amazing video! Thanks a lot for making it.
Hi Aliaksandr,
If you're interested in Hardware and interesting questions asked in the industry, please check us out! ua-cam.com/channels/7h3PROcX7Zgx00alQokJ-w.html
I’m a patron :) I didn’t my get a chance to check in a lot but this is great content. Thank you!
Love the little bubbles on the water analogy diagrams.
What a great tutorial on Transistors! thx
This really taught me a lot. Thank you bro!
Hi,
If you're interested in Hardware and interesting questions asked in the industry, please check us out! ua-cam.com/channels/7h3PROcX7Zgx00alQokJ-w.html
Thanks man :) I will tinker with this ^^
Muchas gracias Moritz!!!
Awesome content!
Hi Moritz, thanks a lot for that video. I'm trying to find out, how the Buchla 100 "Voltage controlled Gate" works. And this explanation gave me access to it.
this amazing you know i have learned a great deal :) and I bet so have you!
Very interesting snd knowledgeful video
Excellent video! Really looking forward to the follow up.
Will you be detailing a current mirror for the transistor emitter bias?
You’re not entirely wrong about the VCR comment. A JFET or MOSFET can be thought of as a voltage controlled resistor, though, when operating in the triode (linear) regions. This is when Vgs-Vt > Vds.
read that often. still not sure if its of any use in synth circuits - haven't seen any JFET filters at all, for example. will look into it!
@@MoritzKlein0considering times when most analog synth circuits were designed, JFET's and other FET transistors were difficult to get , expensive and had wild variances of parameters . Small signal, low noise JFET's are quite in short supply even today. Reason is that it is cheaper to do epitaxial bjt as it makes it kind of "3d" and taking less space, so you can squeeze much more transistors from single wafer, plus epitaxial process is far more predictable, giving you much more stability of yield, that is more transistors with close parameters out of the wafer.
Plus things like cutting, welding electrodes and packaging influence BJT much less than JFET in parameters like capacitance.
Also BJT's can handle much larger current and are much more resistant to static, making them much more attractive and marketable as general purpose transistors.
From hobbyist perspective you can get tons of high quality FET's from things like old tv's, radios, microwave equip like LNA's etc.
Last but not least let's not forget about iron curtain and export restrictions. Most JFET'S could be dual used for radio equipment and exporting and trading them caused risk of Soviet rockets with them raining down the sky, or Soviet KGB getting better bugs which could last longer on a battery. I guess that also did put off designers from putting them into designs - often exported to Soviet union, Italy and other places behind steel curtain.
Very cool, thank you very much! I'll have to put together a couple of these.
For the matched transistors, I happen to have a small supply of TD-101 dual NPNs - if I understand the specs right, they're effectively two matched 2N3904s in the same package, with the pins laid out for easily constructing differential amplifiers. The surplus place I got them from (fairly inexpensively) is sadly sold out, and I'm seeing ebay sellers flogging them off for $6-12 each, so they'd be a terrible thing to build a product for sale around, but they should work fine in this one-off.
yeah looks like those should work well! bummer they're so expensive now.
great tutorial! thanks a lot
Fantastic . Video and channel. :-), great explanation.
14:13 The output curve should be inverted compared to the input curve, since a high input voltage gives a high transistor current, which pulls down the output voltage.
Thank you for mentioning this. It was confusing me as I was working through the video, and glad that someone else was able to confirm this mistake in the comment section!
happened to watch other videos on voltage dividers and voltage followers right before this. quality tutorial, good for you.
terrible to pick but fyi, "regardless" already means what people mean when they say "irregardless" :)
it might be nice, when you do a video to introduce a technology, to also quickly survey where this item fits in in the general development of methods. thank you, lovely work, ought to help a lot of people.
To solve the matching problem just use a THAT-300 transistor array from Mouser electronics (P/N 887-300P14-U) which contains 4 matched npn transistors, sufficient for two separate channels.
will check those out, thanks for the tip!
Hi Michael,
If you're interested in Hardware and interesting questions asked in the industry, please check us out! ua-cam.com/channels/7h3PROcX7Zgx00alQokJ-w.html
you are a huge gift
Very nice video sir.
The analog sound of Kraftwerk :-) BTW: When you use an op-amp with more than one section, it is recommended that you bind the unused + inputs to ground (split supply rails) or to a mid point to the power rail (single rail supply), and to tie the unused outputs to their - inputs, so that the chip remains stable and does not produce instances of oscillations.
Great video! Bravo :-)
I’m curious which envelope generator you were using at the end! Loved the decay on it
What are your thoughts on using something like a THAT 2162 IC instead of matched transistors (aside from the cost difference of course!)
Nice i expected this video a long time! Sehr gut erklärt. I build my vca with a LM13700N its easy but not that funny to bulid its too easy. Good Video
Awesome. 😀👍
Great thing, as always, thank you! Still, I'd like to specify one last detail:
Why can't we bias the transistor up to the normal operating point (>0.7V) and use it in linear amplifying mode? Why do we have to scale the signal all the way down in order to catch that short linear part in the exponential zone instead?
the "linear amplifying mode" is kind of a myth. the relation between base voltage and collector current is always exponential, even at >0.7V. afaik you can really only get it semi-linear by 1) scaling the input down or 2) using an emitter resistor.
@@MoritzKlein0 the kassutronics slope (AD env) circuit has the ability to shape the slope of the attack and decay sections - piecewise it seems like it does more for the least component count and reminds me of my dsp work (xoxos vst) - it's amazing to see someone do really useful stuff with few parts, my understanding of EE is still only far enough to be baffled by how simple some thnigs are and how complex others are.. the slope bending function ought to be availed for all sorts of stuff... this module works as a linear ASR eg as well as the best LFO you could hope to have, i'd like to understand that more, but few people seem to have discovered this functionality.
kassu2000.blogspot.com/2016/04/slope-generator.html
In an amplifier, the linear part of the current curve would be used that lies above 0.7V, but this bias shifting technique to achieve a Voltage Controlled amplifier uses the non linear part of the transistor current curve instead. The non linear region is used to obtain variable gain and, because of this, it is also required that the input be kept small to reduce the distortion caused by the non linear part of the current curve.
@@woodcoast5026 no, again: there is no „linear part“ in the I-V-curve of a transistor
@@MoritzKlein0 OK . The important point is the distinction between a steeper operating region and a flatter operating region.
nice sounds
hello Moritz.
first of all, I want to congratulate you for your channel. I really like it, I'm enjoying each video, and always learn new things. thanks!!
I have a question I've asked in other videos. Do you have any schematic to assemble a symmetric 9v source?
it's a complicated topic and few people talk about it.
I love how neat your videos look. What do you use to make the diagrams appear always in the same place?
From the dots it seems you are using rocket book to create the digital version, but are those actual physical pages on your desk?
nothing digital going on there - i draw everything by hand and then i line the sheets up using the in-camera grid. works surprisingly well!
@@MoritzKlein0 @copernico
They have a 0,5 Millimeter Grid, which helps keeping the same scale of the schematics accross the different sheets.
Hello, Moritz!
I made a VCO like you in your VCO's series and then create VCA like in this video. I plugged them together and connect my Arturia Keystep with Pitch out to VCO and Gate out to VCA. If I don't press any key, I can't hear anything, but if I push a key, I can hear, for example, low frequency note and a nasal high frequency noise on background. How can I clear this noise?
have you managed to fix this problem?
Love these videos, would love to see maybe som videos explain how to buy parts, its super overwhelming looking trying to find the right stuff!
that’s a great idea, will see what i can do!
@@MoritzKlein0 thanks a lot!
I’m in Italy, I’ve tried a few websites, mouser farnel ect… the amount of variations is crazy haha!
Great videos either way, I’ve been hooked just gaining knowledge.
Hi Moritz, what I don't understand from this circuit is that (if I read the PDF from Erica synths) you need to lower the input signal to 20mV peak to peak and add a 570mv offset to get a swinging voltage around 0.6v to get the transistor "working". But if I look at the circuit I only see a 100k/100ohm voltage divider that would lower the voltage, but then it would still swing around 0v. Is there something I'm missing or not understanding in the circuit? Because this way the base would be "pulling" when there is a negative voltage right?
Moritz, I'm still waiting for diode ringmod and op-amp wavefolder videos!
A tutorial on designing an LFO would be a good companion to this video. Perhaps you already posted it, and I blinked and missed it. Op-AMPS or transistors could be used, but I prefer an old IC from the early '80s: the GAP-001.
I'm confused by the breadboard's power rails, which I believe is supposed to be +12V and GND, but I don't see you plug in a power source before your oscilloscope to confirm I am wiring this properly. Unless there is a magical property of your oscilloscope I am unaware of? Sorry, prob a noob question :p
Hi, can you elaborate how does decreasing zero to -2 increase the base emitter voltage? Please...
Always great to see a fresh perspective on circuit analysis. Was confused around 13:30-14:00 why you looked at potential current through the resistor if the transistor wasn’t present, and then look at ratio to actual current in order to calculate voltage at the test point when you could just calculate voltage across the resistor and subtract it from the positive rail voltage. Found that unnecessarily complex. Any particular reason you approach it that way?
Dang this is a year old but I already watched to get your answer lol. I'm not entirely sure why he framed his point at 13:30 the way he did (BJTs are not good VCRs, especially as shown, but they can be configured as a current sinks, sources so it just depends on the application, somewhat irrelevant). Nonetheless, you generally would not know the voltage present at the "output" (the 6 V pt). The output is 6V when biased as shown, but if you cranked the base voltage to a maximum, you would get max current flow and that output voltage would be ~0V (actually around collector-to-emitter minimum voltage 0.3V, so 117uA). Which was his point essentially. That DC output voltage at the collector would fluctuate too much to be considered as a resistor divider.
Great video. It is worth noting that the output on the Common Emitter amp, the voltage on the output would be inverted from the input signal. Your drawings didn't illustrate that.
true!
At the BC548 is a matched Sing what does that mean?
????
whats the song at the beginning?
I guess that the gain will be an exponential function of the control voltage in this case, and which exponential function that is depends on the characteristics of the two transistors. But making the gain dependence on the CV more predictable seems like material for the follow-up video :)
One way to get a more well defined gain should be to use a current source at the common emitter, instead of a voltage source. But then the gain scales linearly with the current.
The voltage gain of a bipolar is approximated by 26/Ie where Ie is in milliamps. 26 is a "constant" derived from the kT/q behavior of the junction. That little equation essentially describes the internal emitter resistor that ratios to whatever resistor is in the collector to set the gain.
In my experience testing it in simulation, it's approximately exponential below about 3 volts, and then linear above that. I've been looking for ways to make it linear all the way down -- the current source is a good idea, thank you!
This video is brilliant. Great job on the explanations. Could I use your circuit and of course mention your channel in a video of mine? I've bee developing a video on VCA's for quite a longe time now...
sure, go for it!
Can we see somewhere the live on easyEDA?
Is the opamp on a bipolar supply? I'm trying to understand how you put out a negative 2V. I assume that the opamp has a true negative supply, below the ground reference shown in the rest of the schematic.
Where do I start? I just completed my APC, I want more!
Hi! quick question: how do you match transistors? do you mean sth like transistor pairs on a single chip, or do you mean measuring transistors and selecting "close" ones? And when you say BC548 - does it matter whether it is BC548A, BC548B or BC548C? Those differ only in gain (hFE), I think. Many thanks!
You can use matched pair structures , like CA3046 - commonly found in old tv's or radios. They are not really great transistors so you can get much more modern structure with lower noise etc. , depends where you live. Point is such chips have not only matched transistors but they are built on same silicon die, making them thermally coupled.
Ofc You cam also "match" them yourself by using "hfe meter" , many multimeters have HFE measurement option which will be good enough for diy audio...
Then ofc it sounds even more interesting if you deliberately mismatching them 😂
@@piotrcurious1131 Got it, thanks a lot!
What's the power supply of the OpAmp please? Is it dual? 'Cause on LTSPice it distorts like a hell rather than act as VCA. Im' trying to figure out the reasons
yes, it‘s +/-12V
@@MoritzKlein0 great.I simulated this circuit with a TL081 opamp on LTSPice, and 2N2222A as NPNs but I get a lot of distortion up there, rather than a voltage regulation of the amplitude. The schematics is 44:42 same values.
Can i do a compresor with this?
why is the voltage swinging so high and not centered around zero going negative?
Hello, first thanks a lot for the exeptional quality of your videos!
I'd like to know if there is a website that you recommand for buying basic components to get going
you can either look up what you‘ll need for the circuits you want to build and hit up pages like mouser or farnell - or you could look into the DIY kits i developed together with erica synths.
@@MoritzKlein0
i definitly will look it up, thanks!
would it be a good idea to replace the emitter resistor for a current source?
if you want the CV response to be more linear, yeah.
you're trying to match the beta value of the transistors right? Otherwise it will have slight offset?
I believe you have built a one-half of the Gilbert cell!
Upto 29min:00s every thing was crystal clear. After that didn't understood anything much and I went to cut off region. What does pulling ground below zero fundamentally mean? I always try to understand circuit with electron flow rather the conventional current flow. and in terms of energy. I hope pulling down emitter below zero means simply increasing forward biasing effect of emitter base junction and increases collector current..... Speaking in terms of energy being supplied, increasing positive end 12 to 14 and decreasing ground from 0 to -2 is same ? Since the base at some constant bias voltage (neglecting signal) the electrons coming from the emitter will be pulled towards positive end eventually increasing collector current. Am i getting it wrong?
Love this channel. I'm just getting into DIY electronics with the hopes of building my own synth components like you. Do you, or anyone here, have recommendations for a good starter multimeter that would be good for this type of work and not break the bank?
Uni-t
@@rarevinylgrooves Nice.. Which model do you use?
i've been using a cheap 20€ multimeter for a long, long time. recently upgraded to a fluke, but tbh, the only difference i notice in my day-to-day use is the build quality. measurement accuracy is comparable. so i'd say just go with something affordable, it won't make that big of a difference!
i'm waiting on a DSO188 (
I've also been shopping for an oscilloscope. Mordax's DATA module seemed appealing since it's 4 channel, AND already a euro rack module, which would make one think it's suitable for diy modular synth needs, but can you attach a probe to it? What are your thoughts on DATA?
I wish you make a video on active bias
cool!
Very impressive tutorial! Thanks! Two questions though. What is the purpose of the 1k output resistor, and how does it affect things? And two, why is the waveform not symmetrical at the last stage where you changed the v divider resistors? Was it the same triangle wave yiu were sending in at that first try, and if so why was it more like an N Than a triangle wave? Anyway, even if you don’t answer, thanks. I’m a tech, went to a two year tech school back in ‘89 and did Malvinos approximation, etc, but it’s been so long, getting back into it now and wish you had taught at our school!
here's an explanation for the output resistor: ua-cam.com/video/aGFb7JbTdNU/v-deo.html and about the waveform: i've switched it from triangle to sawtooth because for some reason, my LFO (which i used for the triangle) was acting up and introduced a lot of noise - should have cleared that up in the video!
@@MoritzKlein0 thanks so much for the reply!
The two problems with the "sketchy VCA" were DC offset and low amplification. Wouldn't it be easier to use a high-pass filter and an opamp amplification stage to fix those just like the final stage of the VCO? Then you don't have to worry about matching two transistors and three pairs of resistors.
That is basically decoupling signal, which moritz mentioned. There is the problem of that stabilizing the dc-signal has a delay-time when we change the dc. In osc we don't change dc so the constant load on cap will stay still. In vca we want to modify the dc-offset, so this will be a non optimal solution. If the control voltage has a fast change and the signal is big enough, we momentarily shoot over limits (-12,+12) resulting to a distortion. Not sure though.
Ah, clever use of a differential output to get around the DC biasing issue, but still making a simpler product than a single-balanced diode mixer/VCA.
I'd like to watch a hour video on making a cheese sandwich next.
Won't the output be phase shifted 180?
for the simple common emitter amp, yes. the VCA itself should be in phase with the input.
I don't fully under why you need a transistor emmiter amplifier, wouldn't it possible to use an opamp instead? 🤔
But, isn't it better to put coupling capacitors in the in and out anyway, despite you fixed the offset?
no, because you might want to use the VCA to modulate CV signals.
38:26 didn't you forget to power the upper + power rail? You connected ground but the 4 pin of the Opamps is basically floating instead of being pulled high.
THE DEEP MIND!!!!!!! AAAAAAAAHHHH!!! THE ROLAND!!! SWEET LORD I LOVE YOU!