DynaComp / Ross Compressor - "whistle" and transistor question

[Thomeeque]

 Hi!

This weekend I have build D&R Comp (http://www.geofex.com/PCB_layouts/Layouts/d&rpub.pdf) in the Ross configuration (but without C2, with C11 = 390pF and with Q1-5 = 2N3904). I have two questions:

(1) Is it normal that there is a persistent audible whistling (oscillations around lets say 5kHz) in the resultant sound (not very loud, but definitely louder than white noise added by circuit; It's frequency is slightly altered by SENSITIVITY pot at highest levels)? Are these oscillations what RG calls "control voltage noises" in the d&rpub.pdf (first line on page 2)? But I was not able to get rid of it by tuning 2K trimmer (only at the positions where guitar signal was killed as well). Besides this I'd say the circuit works as it should.

(2) Had somebody chance to compare this circuit using different types of transistors? RG recommends to use 2N5088 (High gain, low noise NPN transistor), but I was not able to get these, so I have used 2N3904. Would use of 2N5088 made a difference? Or other types (e.g. I've seen someone using MPSA18)?

Thanks, Thmq

[SISKO]

I have no whistling on my uinit and i builed it with bc548 and 2n3906 if i remember correctly. If you want some voltage reading just let me know.

[JDoyle]

You shouldn't have a whistle - you have oscillation somewhere in your circuit.

A quick trick for this - you, yourself, in the flesh, are somewhere around 100-200pF of capacitance to ground, depending on the weather, what you ate for breakfast, if you are hungover, etc. Still, it doesn't matter the exact value, your finger is a good way to shunt highs to ground.

So - start poking around the circuit, literally. You should find at least one spot where if you touch it, the oscillation goes away. That is a good place to start looking to see if there are parts awfully close together, or touching...

Move the wires around and see if that changes anything.

Check the cap right after the 3080 - a problem there would cause oscillation, though not as low freq. as you are experiencing, but that doesn't mean it isn't the problem.

Oscillation in stompboxes is normally due to layout issues. And unfortunately can be a pain in the a$$ to solve sometimtes...

The 2N3904s should be fine - the transistor choice in this circuit doesn't matter as much as people make it out to, the input and output transistors matter the most, in terms of noise, but they are buffers so their 'character' or whatever is fairly irrelevant. The high gain the better though in those spots, in general.

Good luck!

Jay Doyle

[Thomeeque]

 Life is full of surprises, it turned to be noise from stabilized wall wart 

It even crossed my mind before, but what confused me was that frequency of oscillations changed when dialing SENSITIVITY pot - it made me believe that cause must be somewhere on-board and exclude PSU from the list of potential offenders (now I explain this behavior as an reaction of PSU to different current consumption which probably changes with different sensitivity - but I did not spent more time on deeper analysis, I was just happy that with different PSU it works flawlessly - I was not planning to use this wall wart in the future, I've use it just accidentally because it was first and nearest one at the time )

Thanks for your feedback (and for calming me down about those transistors as well)!

T.

[Zben3129]

Life is full of surprises, it turned to be noise from stabilized wall wart 

It even crossed my mind before, but what confused me was that frequency of oscillations changed when dialing SENSITIVITY pot - it made me believe that cause must be somewhere on-board and exclude PSU from the list of potential offenders (now I explain this behavior as an reaction of PSU to different current consumption which probably changes with different sensitivity - but I did not spent more time on deeper analysis, I was just happy that with different PSU it works flawlessly - I was not planning to use this wall wart in the future, I've use it just accidentally because it was first and nearest one at the time )

Thanks for your feedback (and for calming me down about those transistors as well)!

T.

This is the reason why I use batteries ALWAYS when I am testing/debugging for the first plug in. Batteries don't cause noise, and 1/100 times a great  PS can for some reason cause noise and give you a royal headache.

Glad its solved

Zach

[JDoyle]

It even crossed my mind before, but what confused me was that frequency of oscillations changed when dialing SENSITIVITY pot...

The Sensitivity pot controls both the range of the change in compression, or the ratio, as well as the standing gain of the circuit as a whole. It literally sets the current through, and out of, the CA3080 OTA. By changing it you were not only adjusting the current through the circuit but the 3080s ability to charge/discharge any capacitances on its output - thus the changing frequency.

The Ross/DynaComp circuit does not suffer any sort of lax oversight in terms of noise - every standard low noise convention should be followed, metal film resistors, short and careful layout and wire placement, good quality caps, etc. I've advocated testing a couple 3080s and auditioning them for noise; I've also found and had better consitency and quality from CA prefix parts vs. LM prefix parts - though that is going to become a moot point as they eventually all dry up - and given that what's left is what's left, your results may vary considerably from mine.

Also - the emitter follower transistors on the input and the output are the important ones in terms of noise in the signal path - definitely try to find low noise, high gain types, if at all possible.

You can also bypass 10uf and up power bypass caps with .01uF ceramics, if you really want to go for broke.

Though, as has been already mentioned, using a battery can help a good bit and make some of the above a tad unnecessary, such as the .01uF bypass-the-bypass caps.

Glad the problem was an easy fix though!

Regards,

Jay Doyle

[Thomeeque]

 If anybody interested (and not regular guest of pictures thread) see http://thmq.mysteria.cz/bonbonek/

JDoyle: Thanks for explanation, it all makes sense - I'll definitely try to get better transistors for buffers occasionally, than I'll replace trimmer by solid resistors. Rest of used parts should be OK, metallic resistors, foil caps, CA prefix Wires are bit long now, but that's because of this candy-box housing, which is just temporary for tests and tweaks - if I will like the circuit, I'll put it into proper box.

T.

[JDoyle]

If anybody interested (and not regular guest of pictures thread) see http://thmq.mysteria.cz/bonbonek/

JDoyle: Thanks for explanation, it all makes sense - I'll definitely try to get better transistors for buffers occasionally, than I'll replace trimmer by solid resistors. Rest of used parts should be OK, metallic resistors, foil caps, CA prefix Wires are bit long now, but that's because of this candy-box housing, which is just temporary for tests and tweaks - if I will like the circuit, I'll put it into proper box.

T.

I may be the minority here but I wouldn't recommend replacing the trim with two resistors - if anything I'd use a multi-turn trim instead. A 5 mV difference in the Vbe of the input transistors of the 3080 will cause a 20% difference in the current balance - that is pretty sensitive. Plus, with drift, I think it is better to be able to reset things if they start to sound a bit thin or 'off'.

My two cents,

Good luck!

Jay Doyle

[Thomeeque]

Thanks Jay, it makes sense again Multi-turn trim it will be

Btw. about SENSITIVITY pot - original MXR uses X470K pot - does not this X means exponential by any chance? When I set it at 12 o'clock on my unit (with linear pot), there is almost no compression yet, the original MXR already squashes a lot (www.youtube.com/watch?v=zZbxpWNM4bc). But I'd say that my unit does not compress as much as DynaComp generally (even at max), so I don't know (could it be because I did use Ross configuration?)..

T.