Dyna Comp Transistor Changes

[Steve Mavronis]

Tonight I'll be changing out the transistors in my late 70's Dyna Comp clone (built with one MPS5172 and four 2SC1849) with five 2N5088. Luckily for me I designed my custom board layout with individual sockets for the transistors to experiment. My main goal is to reduce the inherent noise level that the vintage Dyna Comp has, still using the CA3080 that it's built around. I've read here and elsewhere using the 2N5088 in Dyna Comp/Ross Compressor clones is a popular choice for all five transistors. I'm assuming any differences in current and gain specs won't affect anything else besides the reducing the noise level and increasing the compression range?

[Mark Hammer]

The inherent limitation of the Dynacomp is its reliance on the 3080.  Go here - http://www.oldcrows.net/~patchell/archives/ca3280_2.html - and look at the comparison of the CA3280 and 3080 for an idea of how that limit functions.

The end result is that the input signal has to be sharply attenuated in order not to overdrive the 3080, and then amplified.  In just about ANY audio circuit, amplifying AFTER the signal has passed through a buncha stuff is a recipe for poor S/N ratio.  Unfortunately, when the Dynacomp originally came out, the CA3080 was pretty much all we had.  The underperformance of the 3080 in terms of both noise and distortion characteristics, was why Roland eventually developed their proprietary OTA - the BA662 - and eventually switched over to THAT chips, and why others started using different chips, like the BA6110, LM13600/13700, and CA3094.

Certainly feel free to experiment with other transistors, but it is unlikely you will get the noise to where you want it.  It is simply an inherently noisy design.

[Steve Mavronis]

Yeah I know the CA3080 has it's issues and have to live with some of that when trying to make an authentic sounding vintage pedal clone from the 70's. Even one of the MXR designers admitted that "it didn't have the best signal-to-noise ratio, but it had a really nice attack characteristic." I'm just going off of other pages about a Dyna-Ross Comp clone and people who have tried it (with a CA3080 installed) saying that the 2N5088 transistor choice did result in somewhat less noise. If those other Operational Transconductance Amplifiers use the same pinout as the CA3080, I can experiment with that as well since it's socketed too. In that case, I probably wasn't wise to name my clone using the chip number designation LOL.

[merlinb]

The dynacomp uses quasi-companding above 1kHz to reduce noise in the upper range (which is more noticeable to the ear). This sacrifices headroom, but the argument is that the guitar doesn't produce much above 1kHz, so you don't need as much head room up there. This leaves you with two ways to reduce noise; both sacrifice headroom:

1: Increase the 1M resistor on pin 3. For example, doubling this resistor will give you about 6dB less noise, but 6dB less headroom across the board. 1.5M would give you 3.5dB less noise, 3.5dB less headroom, and so on. Maybe you can get away with less headroom...

2: Increase the 0.01uF cap on pin 3, and the 0.001uF cap on pin 6, by the same amount. This will lower the frequency where companding starts. Doubling them both will lower it to about 500Hz, so at least you retain the original headroom below this point.

You could do a bit of both 1 and 2 of course.

[Steve Mavronis]

I'll save that resistor and cap idea for last or just live with it because might not be worth the trouble in my cramped board:



I based this off of an actual 70's MXR factory schematic. See this thread - http://www.diystompboxes.com/smfforum/index.php?topic=90705.0

Q1 is a MPS5172 and the rest are 2SC1849. I've seen a PCB photo from an old MXR Dyna Comp that had at least two MSP5172 transistors on it. Could all of them been the same? I've wondered why they choose in the schematic to use a different input transistor than the rest and why many clones have all five transistors as 2N5088 instead of the four 2SC1849. Tonight I'll try replacing all except Q1 and then all five to see if I can hear any difference.

[Nocaster Cat]

A Dynacomp I had that sounded awesome had a 2SC1849 for Q1 and MPS5172's for the rest. My experience from testing transistors is the higher the gains the more hi-fi and/or sterile sounding the pedal became. As always YMMV.

[Steve Mavronis]

I just swapped the 2N5088 transistors in and it sounds better, or 'clearer' than before with somewhat less noise at the same settings. By clearer I mean one thing I noticed before was a slight crackle effect at the very end of sustained notes and chords that now is gone. Plus the volume leveling creating the sustain effect is steady, whereas before I noticed a subtle 'ramping' of the sustain volume.  Hope I'm describing this alright.

I got a collection of ten 2N3088 transistors of two brands - two (gold legged) Motorola and eight silver faced ones desginated with an F - that I assume might be Fairchild. I have two identical Dyna Comp clones to end up comparing, except one has the metal can CA3080 (pictured above) and the other almost done being built with a DIP-8 chip CA3080 version. To spit them up evenly between both pedals, I used the Motorola 2N3088 for the Q1 input transistor and the silver faced ones for Q2, Q3, Q4, and Q6. There is no Q5 on the MXR factory schematic.

[Steve Mavronis]

One more transistor comparision question. I put back the original input transistor MPS5172 (leaving the four new 2N5088 output transistors in place) and didn't notice the noise getting any worse. I swapped back and forth a couple times just to make sure. I beleive the MPS5172 is listed as a low noise transistor too. Just curious because like I previously stated, most Dyna/Ross clones use all 2N5088 (even for the input transistor) and I want to stay as close to Dyna Comp specs as possible except when it comes to noise improvement.

The CA3280 that Mark mentioned above is insteresting except I think it only comes in a 16-pin 'dual' package.

[Steve Mavronis]

Okay I know this is probably a stupid exercise but had to verify to myself (never assume the obvious) that the noise level comparison between different sets of transistors installed in a Dyna Comp clone. I hope feeding 'white noise' and measuring that is a valid method. Conclusion: Using 2N5088 transistors instead mitigates noise levels somewhat better.

Test conditions - PC Soundcard Oscilloscope software signal generator white noise fed into Dyna Comp clone on with both knobs maxed and audio captured with Audacity software.

Note - Audacity Frequency Analysis 0 dB reference point = loudest. More negative dB is quieter:

2SC1849 and MPS5172 (script spec) RMS Volume = -30.2 dB


2N5088 (modded) RMS Volume = -31.0 dB