88dB at 120Hz is A LOT of suppression.
As you add more capacitance to the LR8N3 you reduce the gain.
I think you just have to listen and try.
One of the tricks which the aforementioned Janneman suggested to me decades ago was to make two identical amplifiers with different power supply regulators and feed the outputs to a differential amplifier. If there's a difference, well, quod erat demonstratum.
BTW, Jan's newest edition of the T-Reg is quite nice.
As you add more capacitance to the LR8N3 you reduce the gain.
I think you just have to listen and try.
One of the tricks which the aforementioned Janneman suggested to me decades ago was to make two identical amplifiers with different power supply regulators and feed the outputs to a differential amplifier. If there's a difference, well, quod erat demonstratum.
BTW, Jan's newest edition of the T-Reg is quite nice.
Last edited:
Hmmm...
I could put a better LR8 circuit on a perfboard and try it again.
I don't understand enough about tweaking these regulator ICs to troubleshoot problems on my own.
The original circuit had a 33uF cap bypassing its IN pin and only a 1uF cap bypassing the OUT pin. It worked fine that way, but it would shut itself down if I turned off the preamp and then turned it back on 30 seconds later. Then I'd have to wait several minutes for everything to cool down, then it would power up normally again.
What do you think was causing that behavior? Too much capacitance at the IN pins? Too little capacitance at the OUT pins?
If have a raw B+ to the IN pin of +345V and I want +315V at the OUT pin, what would be a safe set of cap values for bypassing IN, ADJ, and OUT? 1uF on each?
Incidentally, the preamp with the CRCRCRCRC PSU is dead quiet.
Shared raw B+ supply is CRCRC. Then each audio PCB has its own RCRC.
I've got it set up 47uF-2.2k-150uF-2.2k-100uF (that's how it ended up after all the fits and starts). Then each audio PCB has 4.7k-100uF-43k-33uF.
It might be old-fashioned and ugly, but it seems to be working.
On the other hand, when the LR8 regs were working, there was more audible hiss in the preamp audio output. Could this have been high frequency noise from the LR8s? Perhaps Vref noise from the unbypassed ADJ pin?
--
I could put a better LR8 circuit on a perfboard and try it again.
I don't understand enough about tweaking these regulator ICs to troubleshoot problems on my own.
The original circuit had a 33uF cap bypassing its IN pin and only a 1uF cap bypassing the OUT pin. It worked fine that way, but it would shut itself down if I turned off the preamp and then turned it back on 30 seconds later. Then I'd have to wait several minutes for everything to cool down, then it would power up normally again.
What do you think was causing that behavior? Too much capacitance at the IN pins? Too little capacitance at the OUT pins?
If have a raw B+ to the IN pin of +345V and I want +315V at the OUT pin, what would be a safe set of cap values for bypassing IN, ADJ, and OUT? 1uF on each?
Incidentally, the preamp with the CRCRCRCRC PSU is dead quiet.
Shared raw B+ supply is CRCRC. Then each audio PCB has its own RCRC.
I've got it set up 47uF-2.2k-150uF-2.2k-100uF (that's how it ended up after all the fits and starts). Then each audio PCB has 4.7k-100uF-43k-33uF.
It might be old-fashioned and ugly, but it seems to be working.
On the other hand, when the LR8 regs were working, there was more audible hiss in the preamp audio output. Could this have been high frequency noise from the LR8s? Perhaps Vref noise from the unbypassed ADJ pin?
--
Last edited:
An interesting calculus here -- you have 10uA forced down the ADJ pin as one constraint/constant then the R1/R2 values. The Vref noise is gained up by 1+ Rf/Rg.
I have to set up the noise measurement kit and measure. Things get mighty tricky when I go over 40V
I don't know, but it could be related to some capacitor that needs a lot of time to discharge or the valves not being entirely cold yet and therefore starting emission too early. The problem with the LR8N3 is that it can only handle 2 mA at 300 V between input and output unless the current pulse is so short it doesn't get a chance to heat up (and the manufacturer doesn't specify how quickly it heats up).
Probably. The noise of a low-power bandgap reference multiplied by 1 + 220 kohm/820 ohm ~= 269.3 is a whole lot of noise.
Last edited:
With these regulators, PSRR and suppression gets less with higher output voltages, because the loop gain goes down. 30V out is very favorable for a 400V reg.
Jan
The intended use of the LR8 is to facilitate start-up of off-line switchers. As such it has very small current and power capability and as found out, it's noise is pretty high. Which is all irrelevant for the intended use.
It is hard to tweak something for a use that has never been considered in the design.
Jan
Decades ago folks were grafting Sulzer/Didden/Jung regulators into high voltage supplies, a matter of wrapping overvoltage / overcurrent protection around them.
Why the current fashion for 3-pin regulators?
All good fortune,
Chris
Why the current fashion for 3-pin regulators?
All good fortune,
Chris
They are useful. I have been experimenting with 317-type chips as pass device in a superreg. Advantages are high current gain and build-in overcurrent, overheat etc. protection. So far not successful due to several factors.
Jan
Jan
I couldn't disagree that a magic 3 terminal box would be handy; I only wonder if the greater performance of designs from decades ago wouldn't be just as useful for top-performance-requirements (poor supply immunity, by nature) vacuum valve supplies as they are for high supply immunity modern monolythic amp supplies.
Sorry, just railing against fashion audio.
All good fortune,
Chris
Sorry, just railing against fashion audio.
All good fortune,
Chris
Luxman cl 32, 34, 40 are 1500...3000 dollars items these days and no regulator inside except two capacitor multipliers on fillament supply while some consider them the best control preamplifiers with phono in the tube world ..Tubes aren't too fussy with anode voltages, they can withstand twice the max anode voltage for short periods of time .Some of them have brown cooked pcb's and still work 50 years later, just need a recap.I can change the anode voltage by 100 volts and have a hard time hearing the difference in some circuits...
Last edited:
OK, updates on this phono stage build and now it's perceived performance...
I thought I had popped the right channel's source-follower MOSFET because its current draw kept creeping up as the sound would go out in that channel. So I replaced the MOSFET, powered it up, measured voltages to confirm all was good, and it was. I let the preamp warm up on the bench for an hour. Voltages still measured fine. So I powered it down and re-attached the PCB to the chassis (brass standoffs). Powered it back up, voltages were fine but then... uh oh.... There it goes again! The problem wasn't the MOSFET after all.
I re-wetted all solder joints. That seemed to fix it. I brought it over to the sound system and plugged it in. Powered it up, let it warm up for an hour. Uh oh again... Crackles, pops in the right channel only.
Unplugged it, took it back to the bench. Thinking the problem might be a dirty contact in the tube socket, I sprayed cramolin in there and inserted/removed/inserted a tube to scrape the contacts a few times. Let it sit for an hour to dry. Powered it up, voltages all good, even with the PCB screwed down onto the chassis. Hope springs eternal.
So I take it to the sound system and hook it up. Let it sit to warm up. All seems good. Then... the right channel goes noisy again. I swapped tubes around and finally found a pair of Telefunken pulls that don't make noise.
I think the right channel tube socket must be dirty or defective. It's a new Belden micalex (the light brown kind). Does that sound like a possibility?
And now back to the OT...
The simple, passive B+ supply is dead quiet. Extremely low hiss and you have to crank the volume up all the way to hear any hum. This is quite good considering the transformers are on the same chassis as the audio circuits. I consider that part a success.
So the question remains -- If I've completely suppressed hum due to ripple, and I have a well-regulated and quiet heater supply, what are the advantages to installing a regulated B+ supply?
1) Lower PSU output impedance (Zout)? Well, if the audio devices have Zout of around 40k to 50k, and the passive PSU's Zout is let's say 100R, that's a huge ratio of audio Z to PSU Z (500:1).
If the regulated PSU has a Zout of 2R, then yes the ratio gets 50X better (25,000:1). But with a high impedance circuit like this, does that do any good? Have we run into the region of diminishing returns?
2) Regulated supply suppresses ripple better? - How low does this need to go? Is there any perceivable advantage in this kind of circuit from reducing ripple on the B+ from let's say 1mV to the microvolt region?
3) Are there other strategies in PSU design -- for this type of circuit -- that make a bigger difference in final performance than passive PSU vs. regulated PSU? '
For the first time, I used a separate transformer for the heater supply, rather than using the 6.3VCT winding on the plate supply transformer. I used a very basic LD1085 series regulator circuit, with 20,000uF at its IN pin, using Schottky diodes for the bridge rectifier on a 12.6VAC secondary. The voltage at the IN pin is almost 17VDC, the output is set to 12.6VDC. It's working well. Could this use of a separate transformer for the heater supply plus DC rectification with good regulation have reduced perceived hum and noise a lot? Perhaps more than passive vs. regulated plate supply?
I'm not trying to make a contest out of this, or prove any kind of point. I just don't know the answer to the above questions, beyond what I'm finding out by building this thing.
The important lesson I've learned so far is that the decoupling between stages of the RIAA circuit must use larger than expected resistor values, on the order of 25% to 35% of the value of the plate load resistors. If you use 100k plate load resistors, use at least a 27k ohm resistor in the RC decoupling network feeding that stage, or risk nasty subsonic resonances from the PSU. A new rule of thumb to remember. (Or would the super-low Zout of a regulated PSU have made this a non-issue?)
The tragicomic sideshow of the right channel audio board's problems seems to be unrelated to the PSUs.
It would be nice if for a 12AX7-based phono stage, all that's really necessary for good sound out of it is a well-filtered PSU with ripple suppressed into the microvolt regions.
There's room in the chassis for a very small B+ regulator PCB. I'll see if I can physically fit Jan's T-reg in there. I've already started on a small Maida reg on perfboard. If I succeed in making a working Maida reg for this, I install it, and it improves the performance/sound, then there ya go.
--
EDIT TO ADD:
Is the advantage of a regulator that it rejects the audio signal from interacting between stages? In other words, better decoupling between stages from the existing RC networks because the raw B+ coming in has such a low impedance? Is ripple rejection just a happy side effect?
I thought I had popped the right channel's source-follower MOSFET because its current draw kept creeping up as the sound would go out in that channel. So I replaced the MOSFET, powered it up, measured voltages to confirm all was good, and it was. I let the preamp warm up on the bench for an hour. Voltages still measured fine. So I powered it down and re-attached the PCB to the chassis (brass standoffs). Powered it back up, voltages were fine but then... uh oh.... There it goes again! The problem wasn't the MOSFET after all.
I re-wetted all solder joints. That seemed to fix it. I brought it over to the sound system and plugged it in. Powered it up, let it warm up for an hour. Uh oh again... Crackles, pops in the right channel only.
Unplugged it, took it back to the bench. Thinking the problem might be a dirty contact in the tube socket, I sprayed cramolin in there and inserted/removed/inserted a tube to scrape the contacts a few times. Let it sit for an hour to dry. Powered it up, voltages all good, even with the PCB screwed down onto the chassis. Hope springs eternal.
So I take it to the sound system and hook it up. Let it sit to warm up. All seems good. Then... the right channel goes noisy again. I swapped tubes around and finally found a pair of Telefunken pulls that don't make noise.
I think the right channel tube socket must be dirty or defective. It's a new Belden micalex (the light brown kind). Does that sound like a possibility?
And now back to the OT...
The simple, passive B+ supply is dead quiet. Extremely low hiss and you have to crank the volume up all the way to hear any hum. This is quite good considering the transformers are on the same chassis as the audio circuits. I consider that part a success.
So the question remains -- If I've completely suppressed hum due to ripple, and I have a well-regulated and quiet heater supply, what are the advantages to installing a regulated B+ supply?
1) Lower PSU output impedance (Zout)? Well, if the audio devices have Zout of around 40k to 50k, and the passive PSU's Zout is let's say 100R, that's a huge ratio of audio Z to PSU Z (500:1).
If the regulated PSU has a Zout of 2R, then yes the ratio gets 50X better (25,000:1). But with a high impedance circuit like this, does that do any good? Have we run into the region of diminishing returns?
2) Regulated supply suppresses ripple better? - How low does this need to go? Is there any perceivable advantage in this kind of circuit from reducing ripple on the B+ from let's say 1mV to the microvolt region?
3) Are there other strategies in PSU design -- for this type of circuit -- that make a bigger difference in final performance than passive PSU vs. regulated PSU? '
For the first time, I used a separate transformer for the heater supply, rather than using the 6.3VCT winding on the plate supply transformer. I used a very basic LD1085 series regulator circuit, with 20,000uF at its IN pin, using Schottky diodes for the bridge rectifier on a 12.6VAC secondary. The voltage at the IN pin is almost 17VDC, the output is set to 12.6VDC. It's working well. Could this use of a separate transformer for the heater supply plus DC rectification with good regulation have reduced perceived hum and noise a lot? Perhaps more than passive vs. regulated plate supply?
I'm not trying to make a contest out of this, or prove any kind of point. I just don't know the answer to the above questions, beyond what I'm finding out by building this thing.
The important lesson I've learned so far is that the decoupling between stages of the RIAA circuit must use larger than expected resistor values, on the order of 25% to 35% of the value of the plate load resistors. If you use 100k plate load resistors, use at least a 27k ohm resistor in the RC decoupling network feeding that stage, or risk nasty subsonic resonances from the PSU. A new rule of thumb to remember. (Or would the super-low Zout of a regulated PSU have made this a non-issue?)
The tragicomic sideshow of the right channel audio board's problems seems to be unrelated to the PSUs.
It would be nice if for a 12AX7-based phono stage, all that's really necessary for good sound out of it is a well-filtered PSU with ripple suppressed into the microvolt regions.
There's room in the chassis for a very small B+ regulator PCB. I'll see if I can physically fit Jan's T-reg in there. I've already started on a small Maida reg on perfboard. If I succeed in making a working Maida reg for this, I install it, and it improves the performance/sound, then there ya go.
--
EDIT TO ADD:
Is the advantage of a regulator that it rejects the audio signal from interacting between stages? In other words, better decoupling between stages from the existing RC networks because the raw B+ coming in has such a low impedance? Is ripple rejection just a happy side effect?
Last edited:
Yeah, there it is! However, I was having problems with subsonic perturbations with the regulators in circuit and working (although not optimally). The subsonic problems were due to the inadequate decoupling post-regulator.
Sorry Jan, I don't understand what you mean.
Are you making a distinction between ripple rejection and ripple reduction? (I've been writing "ripple suppression" meaning "ripple reduction".)
I think it's obvious that if the B+ has <1mV AC ripple riding on the B+ DC, that there won't be audible hum in the audio output coming from the B+.
It is entirely possible to reduce ripple down to a few hundred microvolts using large value capacitors in a passive supply -- especially if you can afford to drop 100VDC or so and your audio circuits only draw 20mA or less in total.
What are the other compelling reasons for insisting on a regulated B+ supply for this kind of circuit?
Hello,
1) Lower PSU output impedance (Zout)? Well, if the audio devices have Zout of around 40k to 50k, and the passive PSU's Zout is let's say 100R, that's a huge ratio of audio Z to PSU Z (500:1).
If the regulated PSU has a Zout of 2R, then yes the ratio gets 50X better (25,000:1). But with a high impedance circuit like this, does that do any good? Have we run into the region of diminishing returns?
This advantage of tube gear compared to solid state was already mentioned by Jean Hiraga in french l'Audiophile magazine a few decades ago.
I would like to add that it is much easier to use chokes with considerable inductance because they wont be that big because their current rating can be small.
Working with a higher voltage on the caps will also increase the amount of energy available.
Greetings, eduard
1) Lower PSU output impedance (Zout)? Well, if the audio devices have Zout of around 40k to 50k, and the passive PSU's Zout is let's say 100R, that's a huge ratio of audio Z to PSU Z (500:1).
If the regulated PSU has a Zout of 2R, then yes the ratio gets 50X better (25,000:1). But with a high impedance circuit like this, does that do any good? Have we run into the region of diminishing returns?
This advantage of tube gear compared to solid state was already mentioned by Jean Hiraga in french l'Audiophile magazine a few decades ago.
I would like to add that it is much easier to use chokes with considerable inductance because they wont be that big because their current rating can be small.
Working with a higher voltage on the caps will also increase the amount of energy available.
Greetings, eduard
Shortly ... you don't need a regulator if you can reduce the ripple at the same level (inaudible) by other means like RC filters , large caps , chokes and so on .
The low output impedance of a regulator is for maintening the voltage stable vs variable current draw ... not very important for audio stuff .
Why are regulators used ? Because are cheap and the ripple is guaranted to be very low .
The low output impedance of a regulator is for maintening the voltage stable vs variable current draw ... not very important for audio stuff .
Why are regulators used ? Because are cheap and the ripple is guaranted to be very low .
Last edited:
I think you mean motorboating and subsonic response peaks (circuit almost motorboating). I mean deep subsonic signals due to mains voltage variations.