That is one touchy subject.
The simulation is very sensitive to the choice of input cap that's why I have a hard time to believe I have it under control.
With an input amplitude of 0.35V LRSpice tells me the THD is 0.017198%.
You find the .asc file (and my choice of BF862 model) in that thread to play for yourself.
The simulation is very sensitive to the choice of input cap that's why I have a hard time to believe I have it under control.
With an input amplitude of 0.35V LRSpice tells me the THD is 0.017198%.
You find the .asc file (and my choice of BF862 model) in that thread to play for yourself.
Last edited:
Importance of offset
I finally got around to checking offset on my M2 last night. It was over - 0.45 Vdc so I had to turn the pot all the way to opposite end to get -0.090 volts. So how important is offset and does it affect sound? I thought I lost a little with the amp after adjusting but it could have been just too much beer.
My current and voltage across the source resistors are spot on.
Jim
I finally got around to checking offset on my M2 last night. It was over - 0.45 Vdc so I had to turn the pot all the way to opposite end to get -0.090 volts. So how important is offset and does it affect sound? I thought I lost a little with the amp after adjusting but it could have been just too much beer.
My current and voltage across the source resistors are spot on.
Jim
Hi all, I need some advice about my power supply
Can I use a 17+17v 450w tranformer? Is this one, polish made and cheap in Spain:
TST450/006 INDEL - Transformer: toroidal | TME - Electronic components
About main capacitors, are 25v rating safe with above trafo? 27000 or 33000uF? My options are:
Cornell Dubillier 380LX
Cornell Dubillier SLPX
United Chemi-Con SMH (same series as Tea-Bag BOM)
United Chemi-Con KMH (105º)
Is there any advantage using separate boxes for supply and amp?
Thank you very much for your help
Can I use a 17+17v 450w tranformer? Is this one, polish made and cheap in Spain:
TST450/006 INDEL - Transformer: toroidal | TME - Electronic components
About main capacitors, are 25v rating safe with above trafo? 27000 or 33000uF? My options are:
Cornell Dubillier 380LX
Cornell Dubillier SLPX
United Chemi-Con SMH (same series as Tea-Bag BOM)
United Chemi-Con KMH (105º)
Is there any advantage using separate boxes for supply and amp?
Thank you very much for your help
Hi all, I need some advice about my power supply
Can I use a 17+17v 450w tranformer? Is this one, polish made and cheap in Spain:
TST450/006 INDEL - Transformer: toroidal | TME - Electronic components
About main capacitors, are 25v rating safe with above trafo? 27000 or 33000uF? My options are:
Cornell Dubillier 380LX
Cornell Dubillier SLPX
United Chemi-Con SMH (same series as Tea-Bag BOM)
United Chemi-Con KMH (105º)
Is there any advantage using separate boxes for supply and amp?
Thank you very much for your help
I had many delays on my custom Toroidal and eventually canceled the order
And then purchased this unit
Transformer-500VA-2x115V-2x18V-1x36V-Sedlbauer- Model RSO-825059
diyAudio server HTTPS page
Last edited:
Negative FeedBack
Hi all,
I am a new comer to this forum
My last audio project was in the 60-70 from last century as an amateur and always be, no formal training in electronics.
My question:
M2 claimed not using Negative Feedback
FIRST WATT M2
I red the M2 circuit:
1) An elegant optocoupler Q5 having a feed back from R13, R14 to switch in an out D1,D2,D3 for governing the thermal runaway property of Q1,Q2.
2) R13, R14 as Series Current Feedback or Series Negative Feedback.
Could any kind soul enlighten my confusion related to Negative Feedback in this particular M2 circuit.
Many gracias
Hi all,
I am a new comer to this forum
My last audio project was in the 60-70 from last century as an amateur and always be, no formal training in electronics.
My question:
M2 claimed not using Negative Feedback
FIRST WATT M2
I red the M2 circuit:
1) An elegant optocoupler Q5 having a feed back from R13, R14 to switch in an out D1,D2,D3 for governing the thermal runaway property of Q1,Q2.
2) R13, R14 as Series Current Feedback or Series Negative Feedback.
Could any kind soul enlighten my confusion related to Negative Feedback in this particular M2 circuit.
Many gracias
autoformer revisted and more fun
Hi there,
after reading the first watt manual as well as parts of this thread, I think the transformer as given by ZM is actually not the type Nelson Pass used. If you take T1 as 600/15k, the autoformer will give sixfold voltage at the output. In the M2 manual the factor is given as five. An autoformer based on 600/10k does that job nicely. Actually this is analogous to the mistake in post# 1343.
Then for more fun (by ZM thought to be spoiled already, see post# 56):
how about an M2 without emitter degenerators? Would give a nice squared law behaviour of the output and hence MORE useable class A power. The optocoupler could take its sense signal from a power supply branch (dedicated to this channel only!) by measuring the drop over R in the CRC filter. How about that? You may want to smoothen this 'sense' voltage a little to make it more DC
enjoy your recordings!
Hi there,
after reading the first watt manual as well as parts of this thread, I think the transformer as given by ZM is actually not the type Nelson Pass used. If you take T1 as 600/15k, the autoformer will give sixfold voltage at the output. In the M2 manual the factor is given as five. An autoformer based on 600/10k does that job nicely. Actually this is analogous to the mistake in post# 1343
.Then for more fun (by ZM thought to be spoiled already, see post# 56):
how about an M2 without emitter degenerators? Would give a nice squared law behaviour of the output and hence MORE useable class A power. The optocoupler could take its sense signal from a power supply branch (dedicated to this channel only!) by measuring the drop over R in the CRC filter. How about that? You may want to smoothen this 'sense' voltage a little to make it more DC
enjoy your recordings!
15K/600 is 25
root of that is 5
autoformer voltage gain (so second coil stacked on top of primary ) is certainly 1+5 ........ which means 1 goes in , 6 goes out
but , you must think of losses in chain (OS is having 0.95V/V , in best case), so that would be lower than 6V/V
however , what's 1 volt up or down , between friends
and 600/15K is certainly the one , at least to my knowledge ...... but I can't remember did Papa exactly specified that here or not ......
regarding other things you mentioned - feel free to try ........ I'm darn sure that Pa voiced it best to his ears ..... and your ears can be different
I'm pretty helpless , when M2 is in question - whatever came to my mind as possible change , it looks as Babelfishing^3 , so at least triple useless
root of that is 5
autoformer voltage gain (so second coil stacked on top of primary ) is certainly 1+5 ........ which means 1 goes in , 6 goes out
but , you must think of losses in chain (OS is having 0.95V/V , in best case), so that would be lower than 6V/V
however , what's 1 volt up or down , between friends
and 600/15K is certainly the one , at least to my knowledge ...... but I can't remember did Papa exactly specified that here or not ......
regarding other things you mentioned - feel free to try ........ I'm darn sure that Pa voiced it best to his ears ..... and your ears can be different
I'm pretty helpless , when M2 is in question - whatever came to my mind as possible change , it looks as Babelfishing^3 , so at least triple useless
The feedback to which you refer is the bias controller, which regulates the
DC bias current in the output stage to a constant value. It has minimal
interaction with the audio signal because
a) It adjusts a common mode voltage
b) It is bypassed with several thousand mirofarads of capacitance.
You could replace it with a small DC voltage source, but you would not see
a performance difference and would spend lots of time adjusting it as the
amplifier warms up or cools down.
Sir, it is a honour for me to read your reply.
Now I do fully understand about "no negative feed back" on signal path on your M2 design, nothing to do with an "automation" for household management even it is mainly relied on negative feed back mechanism from the output.
Kind regards
Pascal
And what about losing the source R's altogether?
Dear mr. Pass,
and what about my suggestion above, to have the bias circuit not knowing anything about the output, NOR the voltage drop across the source resistors? Of course, me being a simple chemist, I hope you forgive me not telling cathode, emitter or source degeneration apart At least I meant the source degenerating resistors (0R47 in the design). The suggestion remains valid and usable imho. What about having the bias circuit taking its input (I called it 'sense') from the voltage drop across the resistor in the CRC filter in the power supply? The source degeneration is no longer needed in that case, and effectively more class A power would be available from roughly the same design.
kind regards,
Mark Bolech
Dear mr. Pass,
and what about my suggestion above, to have the bias circuit not knowing anything about the output, NOR the voltage drop across the source resistors? Of course, me being a simple chemist, I hope you forgive me not telling cathode, emitter or source degeneration apart
At least I meant the source degenerating resistors (0R47 in the design). The suggestion remains valid and usable imho. What about having the bias circuit taking its input (I called it 'sense') from the voltage drop across the resistor in the CRC filter in the power supply? The source degeneration is no longer needed in that case, and effectively more class A power would be available from roughly the same design.kind regards,
Mark Bolech
Last edited:
It's more tricky than you might think.
Let's take the example of reading the bias current from the voltage drop
across the resistor in the CRC filter. For quiescent bias without current
flowing through the load, you can use the DC voltage to adjust the bias
voltage across the Gates of the outputs. However when significant amounts
of current start flowing through the load and the output starts approaching
the Class A output limit, the voltages across those resistors increase on
average, shutting the bias down.
I have attached a copy of an old patent as an example, where it works on
the driver transistors because they don't have enough variation in their
output current to create this problem. This phenomenon in also part of
the A40 amplifier, where the amplifier would lower the bias when the output
exceeded Class A.
In the M2, this is avoided by a more clever scheme where the drive to the
opto-isolater is limited by the LM385's which coincidentally are very close
to the forward drop of the LEDs.
In any case, any adjustments made by the circuits to the bias is
differential to the bias value, but common mode to the signal - any
noise or distortion it might create is sums to zero at the Sources of the
output devices. And, as I indicated, the circuit is shunted by 3,300 uF
so as to filter any such noise or distortion.
You will get the sense that I am still happy with the circuit, even though I
no longer use it.
Let's take the example of reading the bias current from the voltage drop
across the resistor in the CRC filter. For quiescent bias without current
flowing through the load, you can use the DC voltage to adjust the bias
voltage across the Gates of the outputs. However when significant amounts
of current start flowing through the load and the output starts approaching
the Class A output limit, the voltages across those resistors increase on
average, shutting the bias down.
I have attached a copy of an old patent as an example, where it works on
the driver transistors because they don't have enough variation in their
output current to create this problem. This phenomenon in also part of
the A40 amplifier, where the amplifier would lower the bias when the output
exceeded Class A.
In the M2, this is avoided by a more clever scheme where the drive to the
opto-isolater is limited by the LM385's which coincidentally are very close
to the forward drop of the LEDs.
In any case, any adjustments made by the circuits to the bias is
differential to the bias value, but common mode to the signal - any
noise or distortion it might create is sums to zero at the Sources of the
output devices. And, as I indicated, the circuit is shunted by 3,300 uF
so as to filter any such noise or distortion.
You will get the sense that I am still happy with the circuit, even though I
no longer use it.
Attachments
Thank you for your insights! I downloaded your patent and like the engineering solution. I already felt coming that implementing the control of the quiescent current from taking the voltage drop over the power supply filter resistor is not straightforward . Still, I feel that a comparably elegant solution as you designed into the M2 with the LM385's bootstrapping/limiting the voltage input into the optocoupler, must be achievable in a modified amplifier design without source degenerators. And -quite likely- with more class A power output available at equal quiescent power loss. Possibly a solution 'sensing' the voltage drops over the resistors in the CRC filters of both + and - supply, and using two optocouplers in parallel to define the bias voltage.. Anyway, i will do some more thinking and hope to come up with a solution sometime soon
. Still, I feel that a comparably elegant solution as you designed into the M2 with the LM385's bootstrapping/limiting the voltage input into the optocoupler, must be achievable in a modified amplifier design without source degenerators. And -quite likely- with more class A power output available at equal quiescent power loss. Possibly a solution 'sensing' the voltage drops over the resistors in the CRC filters of both + and - supply, and using two optocouplers in parallel to define the bias voltage.. Anyway, i will do some more thinking and hope to come up with a solution sometime soon