Welcome to Solid State Guitar Amp Forum | DIY Guitar Amplifiers. Please login or sign up.

March 28, 2024, 06:06:41 PM

Login with username, password and session length

Recent Posts

 

Triode emulation with bipolars

Started by aquin43, September 16, 2016, 09:23:02 AM

Previous topic - Next topic

aquin43

Hello all,

I have been dabbling with emulating a triode stage but using BJTs rather than FETs, which I find to be too variable in their parameters.

I have put a report on my website at http://www.aquinaudio.co.uk/valve_em.html and I am posting here, where I hope it may be of use, rather than just being lost in the background noise of the www.

Arthur

J M Fahey


Enzo

Why re-invent the wheel?  Many are doing such experiments.  I suggest looking at any of the Peavey Transtube series of amp circuits.

aquin43

Quote from: Enzo on September 16, 2016, 05:36:10 PM
Why re-invent the wheel?  Many are doing such experiments.  I suggest looking at any of the Peavey Transtube series of amp circuits.

Hi. Its more fun to do one's own thing.

The Peavey stuff seems to be mostly concerned with behaviour at the extremes. Even their darlington input stage has an odd change of harmonic structure with level at lower levels which is far from triode like.

I was trying to point out that a bipolar transistor can behave much like a fet if properly driven and can provide triode-like behaviour where the level of harmonics falls uniformly with falling level. I also provide some theoretical reasoning and references which I hoped would be useful.

The circuit given is completely reproducible with standard parts and no selection or trimming. In production, parts are cheap but time is expensive.

Arthur

Loudthud

#4
The Peavey Transtube does have the raising distortion characteristic, but THD only reaches about 3% before clipping. At low levels I didn't see the 2nd harmonic dominance, 3rd was always present at nearly the same level. It does sound kind of tube like, but not as warm as a real tube or my Bender preamp.

Link to Bender preamp: http://www.ssguitar.com/index.php?topic=3732.0

aquin43

The Transtube seems to act more like a soft clipper acting at a low level. The output is also at high impedance, straight from the collectors. Amusingly, their schematic contains awful warnings about trade secrets and patents. Patenting the Darlington Pair?

On the other hand, the waveforms from the Bender seem much more like the real thing but I couldn't locate the schematic. Can you direct me?

Arthur

g1

Quote from: aquin43 on September 19, 2016, 04:10:31 AM
the waveforms from the Bender seem much more like the real thing but I couldn't locate the schematic. Can you direct me?
The schematic is attached to the first post in that thread
(http://www.ssguitar.com/index.php?topic=3732.msg29076#msg29076)

If you are not logged in you will not be able to see it.

Loudthud

I built the preamp as described in the linked site. I even found a 2N4401 to use for Q1. Note that the Emitter of Q1 is about -0.5V so C2 needs it's + terminal to ground. The only thing I changed was an added diode from the input of U1 to Vee for protection.

It works pretty much as described. The one problem I found is there is a DC shift in operating point at the Collector of Q2 under heavy overdrive. It moves positive about 1.6V. The DC at the Emitter of Q2 also moves around but the Base Voltage is pretty stable. The Emitter of Q4 is also stable. This is not a problem when just a guitar is plugged in.

The DC at the Collector of Q2 is a little high to begin with causing premature clipping on the positive side of the output. An adjustment to R12 would probably fix this. THD was about 7% at the onset of clipping. At lower levels 2nd harmonic is dominate.

aquin43

Quote from: Loudthud on September 21, 2016, 08:47:20 PM
I built the preamp as described in the linked site ...

Thank you very much for your practical evaluation of the circuit.

To address the problems you discovered.

The design was loosely based on one of the standard ECC83 cathode biased setups from the Mullard datasheet. 250V and 47k Ra. I think that I set the voltage at start of grid current too high, causing clipping to start by running out of current as you observed. Moving the cathode of D2 down to the anode of D4 would give a more realistic result. Biasing D1 and D2 from a separate potential divider would give more design freedom.

I am not totally happy with the set up for R10. It should really be fed by a fixed offset from Q2 emitter. The present arrangement allows separate control of the subtracted current via R9 and the curve softening by R10. Unfortunately, the voltage across C5 does vary with signal and a few millivolts at this point is significant.

Extreme overdrive conditions would normally only occur in the second stage or later where the input to the stage would be AC coupled. The behaviour under these conditions is quite different.

I saw that you had plotted an overall transfer function for the Bender, so I thought a similar curve for this circuit might be of interest. This is for the version with earlier grid current onset and includes the 33k in series with the grid.

Arthur


Loudthud

What I was seeing when looking at the transfer function on the X-Y scope was the whole curve would move up about 1.6V depending on the amount of overdrive. The point where hard clipping occurs on the top of the trace would move to the right. You can see the waveform moving around in a normal Y-t display.

aquin43

I think that there is a problem with using the voltage across C5 as the reference for the clipping offset current. Its OK with modest amounts of clipping but moves with heavy overdrive. I have a new version of the circuit that provides a static reference for the offset current. I have also adjusted the operating point to make positive and negative clipping more equal.

Of course, the choice of operating point has a profound effect on the harmonic spectrum as it enters clipping. If it is perfectly balanced, the second harmonic dips sharply and the third harmonic rises to be greater than it. This will happen with a real valve as well. The original version of the circuit was arranged to clip first by running out of current, which seemed to be the behaviour of one of the setups on the Mullard datasheet.

I will post the new circuit tomorrow (UK time) when I have done some more testing.

Arthur

aquin43

I have been reconsidering some aspects of the design. The level shift you observed is caused by the change in the voltage across C5. It was a mistake to try to avoid providing a fixed reference. Its OK when the circuit is being used as a "sweetener" but is not suitable for heavy overdrive. A simple modification solves this problem by providing a fixed voltage across R10.

The two gifs main_dc and main_dc_orig show two variants with fixed reference. They enter clipping with different ratios of H2 to H3 because of the choice of operating point.

Further investigation has brought up the question question of whether emulating the rapid turnoff of the valve is absolutely necessary. The transistor turns off very gradually, following the exponential curve at roughly 10 times reduction in collector current for every 60mv of base drive. In some ways this is an advantage, softening the corners of the clipped waveform. The small amount of feedback sharpens it up slightly.

With this in mind, the third gif, main_dc_noclip shows a much simpler circuit without the dc offset. The circuit is biased to clip symmetrically so that in the clipping region, the second harmonic falls as the third harmonic rises. Diode D3 prevents Q1 from saturating on extreme overdrive, which would otherwise give a non-flat shape to the bottom clipping region.

Arthur

aquin43

I have added a new page to the website describing a much simplified version of the circuit. It doesn't use the separate cascode output clipper and generally has input clipping levels that are half of those of an ECC83. It is also bipolar throughout and can accept inputs exceeding the power rails. This version is intended for use as a gain stage inside an amplifier where the reduced levels are more suited to the lower voltage rails used with transistors. It is also approximately temperature compensated.
http://www.aquinaudio.co.uk/valve_em_further.html

Arthur