Thanks Greg,
I think that might be instability with my amp.
I have checked with my basic scope, the amp was not clipping with my input signal (1.3V p-p for sine wave at -1dB when the volume pot is at max). My Fetzilla can output 55V p-p square wave with my simple square wave generator. I used an Zobel (10R+47n) on my setup, the Zobel resistor - R17 got hot when it was doing 55V 16kHz square wave.
I am using a 200R for the VAS gate-stopper. I will increase that to 470R & I may add a small miller in the next week or so.
Cheers, Stanley
Well, it can be done.
I have several amps fully stable with no inductor, and it is simple to fix.
For BJT you need base resistor (10R) and emitter should be no less than 0.22R.
If the outputstage is a follower ad a cap from the output to the VAS. It will only influence the amp at high frequency or when the gain start falling from ~1 to less.
I have a problem.
Yesterday I switched to a higher output bias and checked the output offset.
Switched off and powered up an hour or so later.
The output offset started as normal, big transient peak and then settles very quickly to a highish offset of +40mV. I note that the offset starts to increase rather than the usual decrease and seems to accelerate away. Switch off rapidly. What did I do wrong?
Today I checked the circuit and the voltages and the currents and decided I had not done anything wrong, nor that anything had been damaged.
So back to monitoring at the higher bias currents. Went up to 350mA OK, up to 400mA OK, and finally up to 450mA OK. Let it soak at this for a while and temperatures came up and output offset holds near zero with a little bit of trimming. As usual it starts with a high +ve output offset and moves towards negative as the circuit warms up.
Switched off to let it cool. Switched on and a second time the offset ran away. I let it rise from ~+55mV to +300mV and switched off.
It appears that there is a basic instability of the mechanism that sets the output currents and thus output offset when aiming for higher ClassA bias.
I did not see this behaviour when testing between 0mA to 250mA of output offset.
Could the Lateral mosFETs need more stability enhancing tweaks other than their inherent -ve temp co once above ~100mA?
Yesterday I switched to a higher output bias and checked the output offset.
Switched off and powered up an hour or so later.
The output offset started as normal, big transient peak and then settles very quickly to a highish offset of +40mV. I note that the offset starts to increase rather than the usual decrease and seems to accelerate away. Switch off rapidly. What did I do wrong?
Today I checked the circuit and the voltages and the currents and decided I had not done anything wrong, nor that anything had been damaged.
So back to monitoring at the higher bias currents. Went up to 350mA OK, up to 400mA OK, and finally up to 450mA OK. Let it soak at this for a while and temperatures came up and output offset holds near zero with a little bit of trimming. As usual it starts with a high +ve output offset and moves towards negative as the circuit warms up.
Switched off to let it cool. Switched on and a second time the offset ran away. I let it rise from ~+55mV to +300mV and switched off.
It appears that there is a basic instability of the mechanism that sets the output currents and thus output offset when aiming for higher ClassA bias.
I did not see this behaviour when testing between 0mA to 250mA of output offset.
Could the Lateral mosFETs need more stability enhancing tweaks other than their inherent -ve temp co once above ~100mA?
Last edited:
AndrewT,
I am really confused by your issue, it may be due to your inverted topology. Please see post #710 for a plot of my dc offset with the bootstrap ccs.:
http://www.diyaudio.com/forums/soli...as-lateral-output-perfect-71.html#post2594392
By the way, your bass will be 10x better with a ccs in the vas.
Cheers,
Greg.
I am really confused by your issue, it may be due to your inverted topology. Please see post #710 for a plot of my dc offset with the bootstrap ccs.:
http://www.diyaudio.com/forums/soli...as-lateral-output-perfect-71.html#post2594392
By the way, your bass will be 10x better with a ccs in the vas.
Cheers,
Greg.
I opted to try the simple 4transistor version first.
I could insert a jFET to replace R8 in post1404.
Note, that I have chosen not to use source resistors in the output pair.
What is the worst case voltage across R8 when it is a CCS? During what operational mode?
Your link took me to post1401 ????
post710 graph shows what I usually find at start up. A big pulse followed fairly quickly with a high offset voltage that reduces as the stage warms up.
I do not get a sudden change to high offset. It settles to the high offset and then instead of going lower it goes higher and drift gets faster as the stage warms up in that first 30 seconds or so.
I could insert a jFET to replace R8 in post1404.
Note, that I have chosen not to use source resistors in the output pair.
What is the worst case voltage across R8 when it is a CCS? During what operational mode?
Your link took me to post1401 ????
post710 graph shows what I usually find at start up. A big pulse followed fairly quickly with a high offset voltage that reduces as the stage warms up.
I do not get a sudden change to high offset. It settles to the high offset and then instead of going lower it goes higher and drift gets faster as the stage warms up in that first 30 seconds or so.
Last edited:
Well I'm sure they said different things at different times but I read from John Curl that you should design so as to avoid the need for o/p chokes if you possibly can but if you really can't manage that then at least keep the inductor size down as much as possible. He was quite clear in his opinion that an o/p inductor spoils the sound. I realise many here don't like to hear this point of view expressed but that does not mean he didn't hold it
The JLH Wireless World published 80-100 Watt amplifier had MF output stage with no CFP and 0.22 o/p resistors plus snubbers - so it looks like he proved himself wrong with his most accomplished design !
mike
Andrew, you may want to include Source resistors for the Outputs. I have also faced a similar rise in offset voltage, when raising bias in an amp with Vertical Mosfets; I found that 100mA worked well in that design and set both channels. Incidentally I also eliminated Source Resistors. I'd also suggest that you try changing the Output Gate resistors to 680E and 470E respectively.
JLH Simple class A, 1969 & JLH 80-100 watt amplifier published in the 1980's - not such a new strategy really.
I wonder do you refer to DNM ?
Unlike Mr Moorecroft I prefer tightly twisted speaker leads for several reasons including reduction of RF pickup but mostly because to me, they just sound better.
mike
p.s by the way add the Nelson Pass Firstwatt F5 to that list
Last edited:
Andrew,
Just saw your post.
I believe latfets do require more stabilisation of bias current...... it is only beyond about 500mA that the tempco becomes negative. Whilst cold in particular the bias creeps up rapidly and can spike before temp brings it down again.
For this reason on my pcb I have used three series 4148 diodes which can be mounted on the reverse side of the pcb, facing the heatsink, giving a thermal feedback mechanism through direct radiation and some convection. I have also added source resistors, and would actually prefer to use 0.22R. This is, I believe, a belts and braces approach.
Cheers,
Hugh
Just saw your post.
I believe latfets do require more stabilisation of bias current...... it is only beyond about 500mA that the tempco becomes negative. Whilst cold in particular the bias creeps up rapidly and can spike before temp brings it down again.
For this reason on my pcb I have used three series 4148 diodes which can be mounted on the reverse side of the pcb, facing the heatsink, giving a thermal feedback mechanism through direct radiation and some convection. I have also added source resistors, and would actually prefer to use 0.22R. This is, I believe, a belts and braces approach.
Cheers,
Hugh
Attachments
Last edited:
Hugh, the indicative figure of 500mA and above to achieve negative tempco is doubtful in practice. Elektor and other designs have routinely used 100mA per output pair for the Hitachi Lateral Mosfets of which I have built too many and have never seen this phenomenon of rising bias or offset.
Sorry Steve, my mistake - it was ETI 1984
http://www.tcaas.btinternet.co.uk/ETI7-84.gif
I have the articles but they are in transit just now but I remember he mentioned that he was pleased that he managed to dispense with the o/p choke and he also, after initial skepticism, agreed that an electrolytic cap in the feedback path spoilt the sound so he designed the FB circuit it so he could use a film cap ( actually two parallel film caps ) instead. The distortion was so low he had to measure it at 10Khz to see anything above the noise floor of his measurement equipment.
This was the amp that was considered one of the best ever designed by those who heard it at the time.
mike
http://www.tcaas.btinternet.co.uk/ETI7-84.gif
I have the articles but they are in transit just now but I remember he mentioned that he was pleased that he managed to dispense with the o/p choke and he also, after initial skepticism, agreed that an electrolytic cap in the feedback path spoilt the sound so he designed the FB circuit it so he could use a film cap ( actually two parallel film caps ) instead. The distortion was so low he had to measure it at 10Khz to see anything above the noise floor of his measurement equipment.
This was the amp that was considered one of the best ever designed by those who heard it at the time.
mike
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.