Bob Cordell Interview: Power Supplies

Status
Not open for further replies.
G.Kleinschmidt said:
Just finished drawing the low current power supply circuit for the voltage amplifier part of a design of mine that requires +/-60V and +/-20V rails.
Click to expand...
Thanks. I've been looking around for a circuit to do regulated rails too, for the same reasons as yours. I just need +/-60V or so, one pair of rails, for simple Blameless-type amps. Such regulators will also be great to power the new NS chips like the LM4702/49811/48930, right?

Till now the only circuit I'd found was the BobEllis one which they had discussed in the context of their active filter board. BobEllis had given details of modifying the component values for 60V rails.
 
I am impressed by the concentraed competence in this thread. I read through most of it last night and it took me several hours.

I am a bit surprised though that we have seen many theoretical debates, but few descriptions how exactly the sound changed in a given circuit.

That is nevertheless my exact question and a bit I believe the original question by BOB...so, would a CLC-arrangement with normal caps outperfrom a CRC arrangement with Black-gates etcetc?

Here is, what my experiences so far are (btw I have build many PSUs for tube amps, for solid-stated, regulated unregulated with 2-poles, with 4-poles etcetc):

Currently I have a nice project playing wonderfull music, the PASS F4-Power amp. It is a Class A- Buffer, so no amplification. It operates at +/-24V and needs approx. 3A per positive and negative rail. It is one of the cleanest, uncolored power amps I ever heard with an incredible 3D-effect. I use it in brideged mode, so two amps using one PSUs. You can get all the infos at www.firstwatt.com

So, I want really this amp to sing and show me what it can deliver in terms of transparency, speed, and especially this "it's simply-there factor", so the feeling that you can touch each tone, here the last vibrations of an instrument.

A. I started with a normal setup: Cheap 300VA-TRansformer from a eölectronic distributor, CRC-Arrangement with 22000uF-TSHA-Panasonic-Caps and non-inductive 0.11-Resistors as the "R". - It sounded very nice, very smooth, a bit lack of transparency maybe, but very, very musical. The 3D efffect was significant, but more on the mid-range/low-trevbles, a bit "air" is missing.

B. I change to a High-End 800VA-Transformer. No Humm, no noise. Something around 1.4T to avoid the noise from industrial-transformers. MU-Shielded, potted. Dead-quiet. Nice. I added at the same time now 4-poles Jensen: 4 pcs per channel, 25V/47000uF each, so we double the capacity. I read in the data-sheet, that the jensens have some kind of build in CLR-Filter, so I withdrew the "R" and uses them first in series, so: Output terminal of one 47000uF goes into the input-terminal of the following one. This should give much better filtering, but as well multiplies the impedance of the two caps.

The sound was a bit solid-state...nice, pronounced trebles, seems to be clearer than the Pnasonics, but as well edginess, a bit metallic sounding. OK, I had the Panasonic now playing for four weeks, maybe a burn-in-effect ?Well, I let them now play non-stop for 5 days...became better, but still....by the way: Rest-Ripple in this setup an impressive 0.3mV !! Much less than hwat PSUDII would recommend.

C. I changed the setup of the JEnsens to parallel them, so each input-terminal goes parallel, each output-terminal as well. Impedance is reduced significantly. Ripples went up to 83mV.

The sound...may be I need to listen more...but m initial reaction is: It is less 3D than either A. or B.. It has by far the best, biggest, blackest bass perfromance. So, it sound warm, not mettallic or technical at all. But it seems to have lost this micro-resolution, the impression, that you can touch each nuance of a played note is gone.

Well....that made me wonder...B and C are only "C"-Setups, no CRC. In all of the cases there is nothing to hear from the amp as noise, it is dead-quiet.

So, the good questions are: What will sound best ? Whcih setups, which cap-types ?

I have many different cap-types on hand to play with:
- Panasonic TSHA
- The 4-pole-Jensen
- Many Panasonic FC (eveb though they have only 2200uf)
- Aerovox
- Philips
- BHC-Slitfoils
- ....Cerafines ? Black Gates ?

I have as well some nice 35mH / 6A-chokes here.

So, what would be your best guess what would sound best ? Currently I am not buying onto the Jensens completely, even though they have theoretcal advantages, my ears are not yet satisfied.

May that changes with a CLC-setup like:
- small Panasonic 2200 or 4400-CAp - 35mH chokes - 94000uF Jensen ?
- or better a CRCLC-setup ? Or the other way round: a CLCRC ?

What would be your best guess (and plese no wise comments like "Simply figure it out", I will , but it would be great to understand what your ears have told you...here are more than a couple of hindreds years experience of amp development present, so I guess we do have listening experiences regarding PSU-setups).
 
Blitz said:
I am impressed by the concentraed competence in this thread. I read through most of it last night and it took me several hours.

I am a bit surprised though that we have seen many theoretical debates, but few descriptions how exactly the sound changed in a given circuit.

That is nevertheless my exact question and a bit I believe the original question by BOB...so, would a CLC-arrangement with normal caps outperfrom a CRC arrangement with Black-gates etcetc?

Here is, what my experiences so far are (btw I have build many PSUs for tube amps, for solid-stated, regulated unregulated with 2-poles, with 4-poles etcetc):

Currently I have a nice project playing wonderfull music, the PASS F4-Power amp. It is a Class A- Buffer, so no amplification. It operates at +/-24V and needs approx. 3A per positive and negative rail. It is one of the cleanest, uncolored power amps I ever heard with an incredible 3D-effect. I use it in brideged mode, so two amps using one PSUs. You can get all the infos at www.firstwatt.com

So, I want really this amp to sing and show me what it can deliver in terms of transparency, speed, and especially this "it's simply-there factor", so the feeling that you can touch each tone, here the last vibrations of an instrument.

A. I started with a normal setup: Cheap 300VA-TRansformer from a eölectronic distributor, CRC-Arrangement with 22000uF-TSHA-Panasonic-Caps and non-inductive 0.11-Resistors as the "R". - It sounded very nice, very smooth, a bit lack of transparency maybe, but very, very musical. The 3D efffect was significant, but more on the mid-range/low-trevbles, a bit "air" is missing.

B. I change to a High-End 800VA-Transformer. No Humm, no noise. Something around 1.4T to avoid the noise from industrial-transformers. MU-Shielded, potted. Dead-quiet. Nice. I added at the same time now 4-poles Jensen: 4 pcs per channel, 25V/47000uF each, so we double the capacity. I read in the data-sheet, that the jensens have some kind of build in CLR-Filter, so I withdrew the "R" and uses them first in series, so: Output terminal of one 47000uF goes into the input-terminal of the following one. This should give much better filtering, but as well multiplies the impedance of the two caps.

The sound was a bit solid-state...nice, pronounced trebles, seems to be clearer than the Pnasonics, but as well edginess, a bit metallic sounding. OK, I had the Panasonic now playing for four weeks, maybe a burn-in-effect ?Well, I let them now play non-stop for 5 days...became better, but still....by the way: Rest-Ripple in this setup an impressive 0.3mV !! Much less than hwat PSUDII would recommend.

C. I changed the setup of the JEnsens to parallel them, so each input-terminal goes parallel, each output-terminal as well. Impedance is reduced significantly. Ripples went up to 83mV.

The sound...may be I need to listen more...but m initial reaction is: It is less 3D than either A. or B.. It has by far the best, biggest, blackest bass perfromance. So, it sound warm, not mettallic or technical at all. But it seems to have lost this micro-resolution, the impression, that you can touch each nuance of a played note is gone.

Well....that made me wonder...B and C are only "C"-Setups, no CRC. In all of the cases there is nothing to hear from the amp as noise, it is dead-quiet.

So, the good questions are: What will sound best ? Whcih setups, which cap-types ?

I have many different cap-types on hand to play with:
- Panasonic TSHA
- The 4-pole-Jensen
- Many Panasonic FC (eveb though they have only 2200uf)
- Aerovox
- Philips
- BHC-Slitfoils
- ....Cerafines ? Black Gates ?

I have as well some nice 35mH / 6A-chokes here.

So, what would be your best guess what would sound best ? Currently I am not buying onto the Jensens completely, even though they have theoretcal advantages, my ears are not yet satisfied.

May that changes with a CLC-setup like:
- small Panasonic 2200 or 4400-CAp - 35mH chokes - 94000uF Jensen ?
- or better a CRCLC-setup ? Or the other way round: a CLCRC ?

What would be your best guess (and plese no wise comments like "Simply figure it out", I will , but it would be great to understand what your ears have told you...here are more than a couple of hindreds years experience of amp development present, so I guess we do have listening experiences regarding PSU-setups).
Click to expand...


Hi Blitz,

Thanks for your interest in this thread, and for sharing your experiences. I can't say that I have a guess, and in fact you have done a lot more experimenting with the different arrangements that you described than I have. Here are just a couple of observations and opinions of mine, but I don't claim to be a power supply expert:

I believe that it is true that sometimes a much higher quality component can make a good difference, but that sometimes as well when multiple standard components are used more wisely they may end up the winner.

Things may differ a great deal in results of power supply management from amplifier design to amplifier design, due to the fact that different amplifier designs may have drastically different power supply rejection characteristics.

It may often be difficult to separate out the effects of power supply components and topologies from effects of different grounding.

In addition to looking at power supply ripple at idle and under load, try to take a look at what the supplies are doing and what is on them when music is being played where the kinds of discernable audible differences you describe are occurring.

Thanks again,
Bob
 
I have to make a correction, my fault: With switching the PSU-Setup of the Jensen, I switched as well the NF-Cable between DAC and Pre-Amp ( from a burned-in WW Golden Eclispe 5 to a brand new Golden Eclipse 5.2). I switched back todoay when I thought how can this difference be...and got back to the sound of B., but even more disciplined, stiffer, sharper ven and very well focussed. So, if you use Jensen's, than in the parralel mode....

Bob, I understand that a PSu will always depend as well on the amp it is playing in...nevertheless I thought, some subjective opions of you guys would be pretty interesting, that is how I learned over the years...sometimes you have the same observations than others and than you know, that this is not pure audio-psychology happening and that you are hunting a ghost...
 
Use of Triacs for mains switch

I'm toying with the idea of using a triac as the mains switch on an amplifier design. I would like to know of any opinions or experiences on this.

In the past I've always used a mechanical mains switch or a relay. In some cases, I've used two relays, one for the main AC current path and one for a parallel soft-start path.

Are there any significant disadvantages to using a Triac?

When using mechanical switches, it is common practice to use a double-pole switch that opens both the hot and neutral mains lines to the transformer. Is this a requirement (e.g., UL)? If it is a requirement and one uses Triacs, then must one somehow arrange to employ two Triacs, one in the hot and one in the neutral?

Thanks,
Bob
 
Check out Nelson Pass's PSU design for the A75.
http://www.passdiy.com/pdf/a75p2.pdf
I never even considered UL requirements when I built this and can only tell you that the end result works great. You can use a very small switch in the front of the amp, where you really want it, and still keep the major current carrying components in the back next to the mains supply and away from the middle of the chassis and signal components.
I have had a supplier's rep mention to me once that solid state switches do bleed current a bit but I'm not too sure how much or to what effect. I assumed they may cause charging of PSU capacitors even in an OFF state but that mattered little to me because I always assume that caps are charged and work from there.
 
Re: Use of Triacs for mains switch

Bob Cordell said:
I'm toying with the idea of using a triac as the mains switch on an amplifier design. I would like to know of any opinions or experiences on this.

In the past I've always used a mechanical mains switch or a relay. In some cases, I've used two relays, one for the main AC current path and one for a parallel soft-start path.

Are there any significant disadvantages to using a Triac?
Click to expand...

Hi Bob,

The disadvantage I'm aware of is that the triac can fail as a short.

The Bryston amps have an interesting triac turn-on facility that also doubles as a soft-start circuit. They combine this with a circuit breaker that is normally on, and not used as the on/off switch per se. An example schematic is in this PDF file. It makes use of a TDA1085C AC motor speed controller chip. Since this chip is designed for washing machines, maybe its use would result in very clean sound :clown:. Dimitri has pointed out that during start-up, the circuit could cause some transformer mechanical hum due to asymmetrical chopping of the AC voltage. This doesn't bother me too much as long as it were only for a few seconds. They use a funky bridge rectifier and large capacitors to take up any DC offset that might appear at the AC line (up to 1.4 Volts), so I don't think persistent mechanical hum would be a problem.

It's a weird circuit that I haven't completely figured out the details of yet. The poorly written datasheet of the TDA1085C does not help this situation either. I intend to implement a similar circuit but have not done so yet. What appeals to me about it is that it should work well when powering up the amp slowly using a variac. The relay-shorted resistor approach seems like it would require a jumper to force the relay closed when powering up the amp with a triac. Otherwise the power resistor could be burned out due to the relay staying open as the variac voltage is slowly increased by hand. I'd be interested in discussing this circuit further with anyone else who is interested, as there are a number of details about it that I still find baffling.
 
Re: Use of Triacs for mains switch

Bob Cordell said:
When using mechanical switches, it is common practice to use a double-pole switch that opens both the hot and neutral mains lines to the transformer. Is this a requirement (e.g., UL)? If it is a requirement and one uses Triacs, then must one somehow arrange to employ two Triacs, one in the hot and one in the neutral?
Click to expand...
Here in Germany we have non-polarized power plugs and still single pole switches are legal for consumer devices... so an apparently switched-off device still can have the "hot" applied on the xformer, etc. I always refused to understand this (given the regulation paranoia in other fields).

- Klaus
 
so an apparently switched-off device still can have the "hot" applied on the xformer
Click to expand...
and the equipment is still safe then :smash:
But I have never seen single pole switches in consumer audio. Is there a special 'lex amplifier' somewhere?
Regards
 
Re: Re: Use of Triacs for mains switch

andy_c said:


Hi Bob,

The disadvantage I'm aware of is that the triac can fail as a short.

The Bryston amps have an interesting triac turn-on facility that also doubles as a soft-start circuit. They combine this with a circuit breaker that is normally on, and not used as the on/off switch per se. An example schematic is in this PDF file. It makes use of a TDA1085C AC motor speed controller chip. Since this chip is designed for washing machines, maybe its use would result in very clean sound :clown:. Dimitri has pointed out that during start-up, the circuit could cause some transformer mechanical hum due to asymmetrical chopping of the AC voltage. This doesn't bother me too much as long as it were only for a few seconds. They use a funky bridge rectifier and large capacitors to take up any DC offset that might appear at the AC line (up to 1.4 Volts), so I don't think persistent mechanical hum would be a problem.

It's a weird circuit that I haven't completely figured out the details of yet. The poorly written datasheet of the TDA1085C does not help this situation either. I intend to implement a similar circuit but have not done so yet. What appeals to me about it is that it should work well when powering up the amp slowly using a variac. The relay-shorted resistor approach seems like it would require a jumper to force the relay closed when powering up the amp with a triac. Otherwise the power resistor could be burned out due to the relay staying open as the variac voltage is slowly increased by hand. I'd be interested in discussing this circuit further with anyone else who is interested, as there are a number of details about it that I still find baffling.
Click to expand...


Hi Andy,

Thanks. That circuit is indeed tough to sort out.

Cheers,
Bob
 
Hi all

An interesting series of articles about decoupling and bypass capacitors, power-supply rails, models, and their interactions can be found here

Regards
George
 
syn08 said:
6kW, you are insane 🙂 You need a foldback protection.
Click to expand...


Hey!

On a different topic:

I'm having difficulty making sense of your TRIAC trigger circuit described here:
http://www.synaesthesia.ca/Auxiliary-circuitry.html

I can’t see why the comparator or the oscillator is required at all. Since you are using a sensitive gate TRIAC with <5mA gate trigger current, all you really need is a bigger C2 and a gate current limiting resistor between the TRIAC gate and the anode of D5 instead of all the other stuff in the middle.

Once the voltage on C2 rises sufficiently above the gate trigger threshold voltage to force the required gate trigger current through the gate series resistor, the TRIAC will fire.
The initial firing will be random phase, but then on the TRIAC will continue to fire on the zero crossings. When a TRIAC is fired (conducting) it ceases to conduct gate current, so the gate conduction duty cycle will be very low and the current consumption next to nil.

Cheers,
Glen
 
G.Kleinschmidt said:



Hey!

On a different topic:

I'm having difficulty making sense of your TRIAC trigger circuit described here:
http://www.synaesthesia.ca/Auxiliary-circuitry.html

I can’t see why the comparator or the oscillator is required at all. Since you are using a sensitive gate TRIAC with <5mA gate trigger current, all you really need is a bigger C2 and a gate current limiting resistor between the TRIAC gate and the anode of D5 instead of all the other stuff in the middle.

Once the voltage on C2 rises sufficiently above the gate trigger threshold voltage to force the required gate trigger current through the gate series resistor, the TRIAC will fire.
The initial firing will be random phase, but then on the TRIAC will continue to fire on the zero crossings. When a TRIAC is fired (conducting) it ceases to conduct gate current, so the gate conduction duty cycle will be very low and the current consumption next to nil.

Cheers,
Glen
Click to expand...

Glen,

A quick one... I'm not sure it's a good idea to fire the triac at every zero crossing. In the current setup, the triac never goes off, that's the whole purpose of the comparator and 10KHz oscillator. The current consumption is very small anyway.
 
syn08 said:


Glen,

A quick one... I'm not sure it's a good idea to fire the triac at every zero crossing. In the current setup, the triac never goes off, that's the whole purpose of the comparator and 10KHz oscillator. The current consumption is very small anyway.
Click to expand...


Hi.

I admit that my thyristor theory is a but rusty. In finalising the PSU circuitry for my current project I needed to do a little revision, but surely the TRIAC will switch of at each polarity inversion and require re-triggering after each zero crossing.
Unless I’m missing something significant, as far as I can see this can be done by simply providing a gate trigger current via a current limiting resistor from a voltage source referenced to the MT2 pin just as effectively as with a free running pulse generator.

Cheers,
Glen
 
G.Kleinschmidt said:



Hi.

I admit that my thyristor theory is a but rusty. In finalising the PSU circuitry for my current project I needed to do a little revision, but surely the TRIAC will switch of at each polarity inversion and require re-triggering after each zero crossing.
Unless I’m missing something significant, as far as I can see this can be done by simply providing a gate trigger current via a current limiting resistor from a voltage source referenced to the MT2 pin just as effectively as with a free running pulse generator.

Cheers,
Glen
Click to expand...

Not if you use the 10KHz (or so) oscillator. This guarantees the triac will be on at at least 100uS before the zero crossing. Such a setup makes the triac to never get off, triacs are very slow devices.

Could you post a proposed schematic? Won't your circuit discharge the capacitor through the gate resistor during the on period?
 
syn08 said:


Not if you use the 10KHz (or so) oscillator. This guarantees the triac will be on at at least 100uS before the zero crossing. Such a setup makes the triac to never get off, triacs are very slow devices.
Click to expand...


When a TRIAC is fired it latches on and stays on until the anode current falls below the holding current IH.
I still don't see how a gate trigger pulse generator can prevent the TRIAC from turning off at the polarity inversion of a 50Hz AC anode current.

Cheers,
Glen
 
G.Kleinschmidt said:

When a TRIAC is fired it latches on and stays on until the anode current falls below the holding current IH.
I still don't see how a gate trigger pulse generator can prevent the TRIAC from turning off at the polarity inversion of a 50Hz AC anode current.
Click to expand...

Triacs and thyristors naturally commutate at the end of each half cycle of load current unless a gate signal is applied to maintain conduction from the beginning of the next half cycle.

To turn off (commutate) a thyristor (or triac), the load current must be smaller than a threshold current for sufficient time to allow a return to the blocking state.

100uS (worst case) is smaller than the time required for the triac to turn off.
 
syn08 said:


Triacs and thyristors naturally commutate at the end of each half cycle of load current unless a gate signal is applied to maintain conduction from the beginning of the next half cycle.
Click to expand...


So why does this gate signal have to come from an oscillator?
Why not just a resistor from a fixed potential of, say, +10V above A2?

Cheers,
Glen
 
Status
Not open for further replies.
Back
Top