Help with VAS CCS stability on blameless amp?

High Cob is problematic for the VAS transistor and it also increases distortion. TO-92 devices generally have low Cob, but look at their datasheet: Below 8pF at VCB=10V is acceptable.
At +/-35V rails and below there are many options for the VAS transistor in TO-92 package.
Above +/-35V but below +/-45V:
TO-92 devices usually have 625mW max. dissipation (must be derated at higher than 25 Celsius deg. temperature), but with your 6-7 mA VAS CCS current (if I calculated right from your circuit), the MPSA42/MPSA92 and 2N5551/2N5401 is still OK for Q7, Q9. and they are widely available. They are not ECB pinwise.
Above +/-45V there's not much choice besides what I recommended in my above post if you look for low Cob and high enough VCEmax and high enough max. dissipation at temperatures that may arise inside amplifier chassis.
R18 at 200 ohms seems too high to me, there's an unnecessarily high voltage drop across it. 20-30 ohms is usually enough. You omitted this resisitor in your 2nd schemtic, but it should be there otherwise the circuit may become prone to oscillation at clipping. Note: the simulation may not show this oscillation.
 
Last edited:
Thank you for the writeup! 🙂 I will keep the rails fairly low. I don't need a lot of power.

I ended up with R18=200ohm since it seemed to damp the transient problems I had. The original PCB had no resistor at all, so I had to cut traces and solder resistors on the back. This is why the schematic from Jim's audio did not have this resistor. I will try lower values with the new transistors.
 
If you look at the datasheets for 2SB649 and 2SD669 you will find these are ideal. Use them and you can run your amplifier with 40 volt supply rails with no concerns.

As genenis69 says the value of R18 is too high and I agree. If you want to degenerate Q9 emitter as per Cordell keep this to about 10%. Keep in mind your circuit is based on Self's original and there is no R18 so be careful with your options.

There have been quite a few departures from Self with some resistor values and I think you should put your trust in his original work with particular interest in the Vas.

The value of R1 should be 2k2. I have suggested this in other threads where lower values have been substituted giving rise to instability.
 
If you look at the datasheets for 2SB649 and 2SD669 you will find these are ideal.

I was looking at using them for drivers, they seem to have higher Ft than the MJE 340/350. In tian-simulation the 0 gain point without miller cap goes from abt 2MHz to 4MHz when I switch to them as drivers (using BC's in the VAS). That seems nice for when I play around with compensation later.

Also, thank you for the other advice. Not sure which R1 you are referring to? I use the designations in my spice-schematic. Q10 emitter resistor?

Another question: on the actual PCB there is a 1N4148 parallel to the miller cap, what is the purpose of that one? Something for clipping behavior?
 
Last edited:
Go ahead as you have planned.

Don't worry about my comment about R1. There should be no diode in parallel with the Miller capacitor C4 in your .asc file.

A diode in parallel with a capacitor can be used to protect an electrolytic type from failure due to a large reverse polarity voltage being applied - possible with C3 if the output goes dc negative under fault conditions.

A parallel diode with the anode connecting to earth would limit the reverse polarity voltage to -0.6V which is survivable. The a.c. signals will be below that level there will be no effect under normal conditions.
 
I was looking at using them for drivers, they seem to have higher Ft than the MJE 340/350. In tian-simulation the 0 gain point without miller cap goes from abt 2MHz to 4MHz when I switch to them as drivers (using BC's in the VAS). That seems nice for when I play around with compensation later.

Also, thank you for the other advice. Not sure which R1 you are referring to? I use the designations in my spice-schematic. Q10 emitter resistor?

Another question: on the actual PCB there is a 1N4148 parallel to the miller cap, what is the purpose of that one? Something for clipping behavior?

Did you get it running to your satisfaction? I wanted to get the Trimodal PCB, but nobody from Signal Transfer Company is answering.

I'm also a bit puzzled on how I should wire up two Blameless/Trimodal with fully balanced inputs. I have XLR pre-outs on my pre-amp I would like to use.