Opamp in theremin

[i]"The better power supply may be a basic requirement for improving the stability and sound quality of the Theremax. But it will not solve all problems. I'm still making tests how I divide best the circuitry which uses this mid voltage by two and decouple the parts." - Thierry [/i]

I agree, Thierry - Splitting the Mid-V rails is likely to improve matters , and may be a simpler way of implementing sufficient improvement so that a seperate active supply / Mid-V regulator is not required.. However, the Mid-V on the Theremax at present (as far as I can see from my almost unreadable schematic) consists of a 1k pull-up and a 1k pull-down.. This is quite a high Z, and lowering this Z down to (say) 47R (or even lower, say 10R) and having a bigger decoupling capacitor (say 1u) would probably obliterate any fluctuations in the Mid-V even if the 'parts' are not seperated.

My thoughts are that, if one is building / fitting a supply regulator anyway, one may as well incorperate an active Mid-V supply into the design.. One could (if one needed to) take this low-z mid-V via a pair of resistors to the respective (now disconnected) seperate parts, a 100R from the Mid-V amplifier output to one 'part' with a 470nF capacitor to ground, and a 100R from the Mid-V amplifier output to the other 'part' with a 470nF capacitor to ground, for example.

But, as I said in my last posting - There is no such thing as a universal PSU scheme.. There are a set of 'rules' which are rarely followed (and from my limited exposure to the Theremax seem to have been completely violated) but which should have been carefully designed in.

I am sure you have a lot more expierience with the Theremax (I have only briefly examined the schematic, have never seen one 'in the flesh', and it is one of the few theremins whose circuitry I have not bothered to build or even simulate) So I believe people would be wiser to heed your advice than listen to mine.

I merely posted my circuit because I had it - I DO NOT vouch that this circuit is the best, or even a suitable solution to the Theremax 'problems' ... I can not even say with certainty that I know what the 'problems' are! LOL!

;-)

Fred.

Any luck with testing, Thierry?
Can swapping 100pF, 68pF NP0 capacitors for Wima polypropylene FKP2 make an improvement? http://www.wima.de/EN/WIMA_FKP_2.pdf http://www.wima.de/EN/article.htm (table at the bottom) You said that whith thicker antennas capacitors, frequency has to be changed. I use aluminium antennas (12mm dia) form the beginning and I measured that I have fairly linear 3 octaves. I can choose which octaves I want to play with pitch trim pot.
I would like to improve lowest notes which are somehow distorted. It seems that FredM's circuit might do it(Thanks Fred). I would build it and try, but since you are doing some tests I'll wait for your ideas and then I'll order components. And as I stated before I would try better opamp in place of old 748, maybe metal foil resistors, better capacitors.

I've built Fred's Split Supply and connected to Theremax in place of old supply. I've split comparator and opamp circuits from rest. 8.2V and 4.1V goes where should. 8.2 is connected through 100R to comparator section with 1uF to ground. Same with opamp section. And there is a problem - sine wave is distorted, positioning timbre pot at sine doesn't give clear sine. I've removed pin 2 - no change. I've removed C30 - got clear sin. Measures: Q8 C=5,57 B=0,6 E=0,02, U1 5=3,84 4=3,84 2=3,56. I also checked voltage on each section, everywhere is ok except comparator where it shows 7.7V.

without C30
[img]http://img522.imageshack.us/img522/8752/screenhunter1502.th.jpg[/img] (http://img522.imageshack.us/i/screenhunter1502.jpg)

with C30
[img]http://img259.imageshack.us/img259/6881/screenhunter1503.th.jpg[/img] (http://img259.imageshack.us/i/screenhunter1503.jpg/)

BTW lows are better now (without C30) than before upgrade

Ok, you say the voltage on the comparator is low.. try shorting out the 100R resistor from the PSU to this section .. If the PSU is built as per my schematic, there is plenty of decoupling, and these 100R resistors should not be required.. they will cause a voltage drop proportional to current drawn - and as the comparator seems to be drawing the most current, the 100R will be dropping the most voltage.

My schematic included 47R resistors - so I suggest bypassing both 100R resistors.

R11 and R12 (47R) on my schematic could be safely reduced down to 10R if more output current / less Vdrop is required.. but it is best to keep these resistors as high as you can get away with - as you approach 10R the chance of instability increases.. And you should leave the local decoupling capacitors (the 1u fitted after the 100r's) on the board

Fred.

Ok, I shorted 100R, now I get 8,2V on comparator. I also have 10R in place of 47R (R11, R12). But it doesn't solve the problem.

Sorry - But I dont know how to help here.

If you are getting clean supplies, then the power supply is probably not the problem. As I said earlier, I know little about the Theremax.. I do not see how the behaviour of C30 can be affected by the change to the supply.. As far as I can see, Q8 buffers the 'raw' AF signal from the mixer, C30 couples this buffered signal to the comparator IC1:A..

Q8:C should be a low-Z output drive.. A sine / sine like waveform should appear here (it is hard to tell from your waveform photos – but even the ones with C30 removed do not look sine to me – they look like ‘rounded top’ square waves – the sort of waveform one gets from overdriving a soft-clipping circuit from a high level sine..)

I suspect the problem is near Q8. If the buffer formed by Q8 is working correctly, then the loading or signal coupling caused by C30, R34 and R43 should have no effect on the waveshape at Q8:C and with the wiper of R81 at the F position, no signal should be coupled to C36 via R40, and a virtually unnattenuated ‘sine’ should be coupled to R36 via C29,R41 and R39 to C36.
Are you sure the wiper is connected? A fault here would certainly give the problems you are seeing!

Fred.

Connecting C30 when timbre is off (without pin 2) causes only slight change and maybe it just loads Q8. Maybe resistors for Q8 aren't correctly calculated (I have 2n3904 in place of 2n4124)and it's load should be lowered(R30)? But I have no knowledge about transistor amplifiers.

I said that removing pin 2 did no change, but I had forgotten that I had removed R51 before so square was so weak that even set on square signal was barely audible, when set on sine square wasn't noticeble.

It seems that there are two separate problems here - loading Q8 and mixing sine with square. When I put R51 back the only solution is to use a jumper in place of R42, then I have to increase R40 beacuse square is too loud and gives uneven transition between sine and squre. I also noticed that my square isn't pure square (sine isn't too lol), look like mix of square and sine, so I put a jumper in place of R41, but it didn't help. [img]http://img207.imageshack.us/img207/5803/screenhunter1505.jpg[/img] (http://img207.imageshack.us/i/screenhunter1505.jpg/)

BTW Vr is now split and goes to separate sections through 10R and 10uF to ground.

Ok, I have just simulated the Q8 amplifier.. 2n4124 and 2n3904 SPICE models give exactly the same results (as expected) and with loadings (C36 and/or C30) fitted or removed there is no significant variation in output (loading into passive loads – ie, resistors to 0V).

Not knowing what the AC signal level from the mixer [D4(A).D5(K).R11.R26] is (and not wishing to simulate the entire circuit or use my brain to calculate.. LOL ;-), I played with various levels and found that any level greater than about 10mV peak (20mV P-P) causes distortion on Q8(C).. This distortion is a result of signal clipping.. It might be worth increasing the value of R26 to reduce the signal level into Q8 a bit.

Voltages seen on simulation (with 20mV P-P input @1kHz) :
Q8(C) = 5.25V to 5.5V ~
C30.R34.R43 = 3.95V to 4.22V ~ (Comparator input)
C29.R41 = -150mV to +150mV (with R41 shorted to 0V by wiper of R81)
Q8:B = 602mV to 619mV

IC1 has open collector outputs – removing R51 will give strange results, particularly if C40 remains in circuit providing an AC path!

Changing the value of R42 is not a good idea.. Reduce square signal level by increasing R40, or by adding a resistor from R39.R40.C36 to 0V (A 470k resistor here will attenuate signals via R40 by 50%, but signals via R39 by only about 5%, a 100k resistor will attenuate signals via R40 to 17.5% but signals via R39 will only be attenuated to 82%) – adding such a resistor should also helps a bit to stabilize the impedance at this junction, particularly at the extreme square setting.

Fred.

[i](Ps - R39.R40.C36 denotes the point where these components connect together.)[/i]

Alas, this is probably as far as I am willing to go with this circuit - I looked at the Theremax schematic several years ago when evaluating designs, and discarded it - there are (IMHO) far too many 'nasty' aspects of the design to make it worth re-engineering (the "waveshaping" circuit which is the present discussion, is one of the most horrible "audio" circuits I have ever seen!).

When I presented the PSU circuit, it was as a general block for general use.. I had (and have) no specific knowledge of the Theremax or suitability of this PSU to it. Looking more closely at the Theremax circuit now, I do NOT think that the original PSU / Vr distribution is likely to be the cause of the problems.. The "problems" are far deeper - and can probably be sunmmed up with one word.. Nasty!

I've put back resitors as they originaly were. I've increased R26 and R59 to 10k. Adding resitors as you said did't help. Later I checked waveform and it's getting even worse - I'm getting spikes! Whatever with or without pin 2, c30, r42. Even when set on square it's squared wave with spikes, but still not a square. It sounds simillar to square, but it isn't a square.
[img]http://img35.imageshack.us/img35/9245/screenhunter1507.th.jpg[/img] (http://img35.imageshack.us/i/screenhunter1507.jpg/) I'll check if lowering/increasing voltage will do anything.

Nasty! Even if it is, it doesn't mean it can't be solved. Maybe I will loose my mind lol, but I won't give up.

[i]"Maybe I will loose my mind lol, but I won't give up."[/i]

LOL! - I know that feeling! ;-)

Perhaps you could tell me what the signal levels (AC) are (at R26.R11.D5.D4) - this may help..

Also, dont be too distracted by the shape of the waveforms.. Remember, there is a frequency dependent phase lag on the signal the comparator actually sees - The square wave will never be exactly in phase with the 'sine' wave, and when the two are mixed it will LOOK like a discontinuous waveform.. but this will look a lot worse than it sounds! In fact, these 'spikes' wont be heard (if they are just the result of phase difference).. Changing anything related to the sine signal will change the point at which the comparator switches, and therefore alter the position of any discontinuities in the mixed waveform.

Waveforms containing exactly the same harmonic spectrum can appear completely and utterly different when viewed on a 'scope .. The reletive phases of the harmonics (which we dont generally hear*) produces this variation in shape, even if the amplitudes of these harmonics are the same [*main time we notice if if the phases combine to produce peaks which cause clipping].. A spectrum analyser is a far better tool for evaluating waveforms generated through signal mixing.


R59 will only affect the volume circuit. Increasing R26 should reduce the signal level into Q8, and if this signal level is too high, should help to inprove the operation of the Q8 buffer.. Try increasing this value to 47k with R42 removed .. The level could be greatly reduced - but you should see a SINE on Q8:C !

Fred.

(PS - get a copy of LT-Spice (http://www.linear.com/designtools/software/).. this circuit is quite easily simulated, and you can play with values and compare what you see with the real board - Its not perfect, but it will indicate the cause of the problem [or give more clues] if there are gross differences between simulation and reality)