hello Rod,
There "is" no problem with the leakage currents. Adding an impedance at the other side of the filament at the cost of more dissipation seems a complex way of solving something that can be solved by layout.
My focus was low voltage drop, high output impedance and low noise. The low noise versions are even 30 dB more quiet compared to the figures I gave earlier in this thread.
nevertheless, keep up the work.
best
one solution is to drill the heatsink 3,8 or 4mm - instead of just 3,3mm for the screw M3.
Then the little Top-Hat pushes into the heatsink, just a little.
How's the filament bias setup, Andy?
Then the little Top-Hat pushes into the heatsink, just a little.
How's the filament bias setup, Andy?
Just so you can gauge interest. I'm also interested in a pp version for my SETH PP2A3 (future build)
However for the CCS part maybe just room for someones own CCS I'm planning to use IXTH-T20N50D cascaded for that part.
However for the CCS part maybe just room for someones own CCS I'm planning to use IXTH-T20N50D cascaded for that part.
I haven't had a chance to critically evaluate all the alternative voltage reg options. I'll have to drill some more holes in the heatsink and work out where to put the boards. Has to be done, though - I need the perfect solution!! I suspect a choke input supply is in there, though.
Andy
Andy
Guido, Let us agree to disagree about leakage current, and common-mode noise!
DIYers can see the diagram I posted, and decide for themselves if there is a problem. If anyone would like to try the difference the Gyrator makes to the sound, the regulator can operate with it bypassed, so a comparison can be made. But I am confident about the improvement it offers. It may add some parts, but these give no difficulties for assembly, and given the very low price of the whole kit, I don't think there can be any dispute about the good value it offers.
The heat dissipated by the gyrator is small, and independent of supply voltage - about 2W for a 300B. This presents little trouble when you mount the regulator on a heatsink, or even the chassis, which is an easy option with the power devices arranged along one PCB side.
Thanks Bas, and John, and Gianluca for your encouragement on the PP design.
I will look again at the PP solution, and see if the costs can be made low enough for a good-value kit.
hello Rod,
If you can show me by calculation that the common mode currents cause a problem, I am the first to agree with you. However, so far I have not seen any calculations from your side. I have to admit that I did not read all pages from this thread, so I may have missed something.
I am not stating your solution does not offer a decent solution, soundwise.
best,
Guido, the only calculation for noise currents would be Ohm's Law.
But to make that calculation you will need to measure the source noise voltage (from mains transformer leakage, safety earth differential-mode noise, rectifier pulses referred to common mode, induction along the cable from other noise sources, etc). You will need the spectrum, as well as the voltage.
Then you will need the effective leakage impedance of the transformer, and all the loop components and conductors, with their effective impedances up to quite high frequencies. As you know, RF intrusion on mains circuits is a real problem, and rectifier pulses can show bandwidth into UHF.
These parameters will vary wildly from place to place, and with every difference in constructional technique and transformer.
The value of such a calculation will vary just as wildly as the configurations you are trying to characterise.
If we really wanted to quantify the problem, then we could set up up an amp in an EMC chamber, and use a wideband differential probe feeding a spectrum analyser to give some ideas about the magnitude of the currents. Still, because of the variability of the problem, I doubt the value of the exercise.
In my day job, perhaps in yours too, I solve problems of conducted and radiated EMI in mass produced equipment. By means of creating stops ( wide bandwidth series impedance) and diversions (shunts) unwanted noise currents can be tamed. I have used a similar approach in the DHT regulators, and the outcome seems to be satisfying to a large number of Amp makers who have tried them.
I believe that one reason for this success is that real amps are built with imperfect components, and imperfect layout. It is OK for our personal amps to eliminate every such imperfection, but we have no right to expect everyone to be able to do this every time. After all, one good transformer costs much more than my regulators, and builders like to use the ones they already have, in many cases.
But to make that calculation you will need to measure the source noise voltage (from mains transformer leakage, safety earth differential-mode noise, rectifier pulses referred to common mode, induction along the cable from other noise sources, etc). You will need the spectrum, as well as the voltage.
Then you will need the effective leakage impedance of the transformer, and all the loop components and conductors, with their effective impedances up to quite high frequencies. As you know, RF intrusion on mains circuits is a real problem, and rectifier pulses can show bandwidth into UHF.
These parameters will vary wildly from place to place, and with every difference in constructional technique and transformer.
The value of such a calculation will vary just as wildly as the configurations you are trying to characterise.
If we really wanted to quantify the problem, then we could set up up an amp in an EMC chamber, and use a wideband differential probe feeding a spectrum analyser to give some ideas about the magnitude of the currents. Still, because of the variability of the problem, I doubt the value of the exercise.
In my day job, perhaps in yours too, I solve problems of conducted and radiated EMI in mass produced equipment. By means of creating stops ( wide bandwidth series impedance) and diversions (shunts) unwanted noise currents can be tamed. I have used a similar approach in the DHT regulators, and the outcome seems to be satisfying to a large number of Amp makers who have tried them.
I believe that one reason for this success is that real amps are built with imperfect components, and imperfect layout. It is OK for our personal amps to eliminate every such imperfection, but we have no right to expect everyone to be able to do this every time. After all, one good transformer costs much more than my regulators, and builders like to use the ones they already have, in many cases.
hello Rod,
I've spent quite some time dealing with EMC in my life (and still do). Assumptions without proof will not bring us anywhere. Let us apply ohms law, as you propose;
Assume a shunt capacitance of 1nF in the mains transformer (primary to secondary). Assume it is fully asymmetrically divided (never the case, but we need a starting point, and this is very worst case). So that 1nF is fully excited by 230V. This yields a current of 75uA. This in turn generates a voltage across a (say) 10 ohm cathode resistor of 0,75mV. Amplified by mu (300b) this gives 3 mV at the anode. This voltage is divided down by 30 (the primary to secondary ratio oif the output transformer) hence yields 100uV at the speaker. This is 80dB down compared to 1V output. Nothing to worry about, moreover, other non parasitic noise often dominates the output noise. My practice shows an average level of 1mV of hum at most commercially available 300b amps.
best
OK Guido, Let's take your 1nF figure for leakage and apply it to a 26 linestage.
With the same 10R cathode resistor, the 0,75mV is applied to the cathode, just as if it were a grounded-grid stage. So, the noise has the same effect as if it were at the grid. For a 300mV source, the hum is only about -50dB down at full volume, and at low volume it could easily be only -40dB down! This is very bad performance, and one reason DIYers complain of hum with the #26, even after using rudimentary CCS heating with LM108x, etc.
Things get even worse for high frequency noise. Some full-AM intrusion at 1MHz, 10mV will appear at -64dB if the cathode circuit is purely resistive. With low volume, and a wirewound resistor (very common choice for a cathode resistor) the impedance rises, and with the degraded performance at low volume, you could be looking at -45dB or so. And this is not hum, it is noise modulated by audio or other noise noise that may be demodulated elsewhere in the amp.
With only 5mV of UHF, say 100MHz, which may be from transmitters, ADSL coupling, rectifier noise, light ballast controllers, switching artefacts from electronic equipment elsewhere on the mains - all of which are very visible on a scope, even in the semi urban environment where I live. Anyway 5mV of 100MHz will break though to the tune of about 4.3mV or -36dB down from 300mV with worse performance in practice, since the cathode resistor, and wiring, show appreciable impedance at high frequencies like these.
I believe that this is something to worry about very much!
The buffering presented by my 26 regulator, with an NPN transistor operating with about
35pF working capacitance, gives a 30dB improvement on these figures - which is easily borne out by the improvement in the sound quality, compared to having no gyrator.
With the same 10R cathode resistor, the 0,75mV is applied to the cathode, just as if it were a grounded-grid stage. So, the noise has the same effect as if it were at the grid. For a 300mV source, the hum is only about -50dB down at full volume, and at low volume it could easily be only -40dB down! This is very bad performance, and one reason DIYers complain of hum with the #26, even after using rudimentary CCS heating with LM108x, etc.
Things get even worse for high frequency noise. Some full-AM intrusion at 1MHz, 10mV will appear at -64dB if the cathode circuit is purely resistive. With low volume, and a wirewound resistor (very common choice for a cathode resistor) the impedance rises, and with the degraded performance at low volume, you could be looking at -45dB or so. And this is not hum, it is noise modulated by audio or other noise noise that may be demodulated elsewhere in the amp.
With only 5mV of UHF, say 100MHz, which may be from transmitters, ADSL coupling, rectifier noise, light ballast controllers, switching artefacts from electronic equipment elsewhere on the mains - all of which are very visible on a scope, even in the semi urban environment where I live. Anyway 5mV of 100MHz will break though to the tune of about 4.3mV or -36dB down from 300mV with worse performance in practice, since the cathode resistor, and wiring, show appreciable impedance at high frequencies like these.
I believe that this is something to worry about very much!
The buffering presented by my 26 regulator, with an NPN transistor operating with about
35pF working capacitance, gives a 30dB improvement on these figures - which is easily borne out by the improvement in the sound quality, compared to having no gyrator.
hello Rod,
Obviously at pre amps things get worse. However, one can always apply capacitors (1uF) between (both ends of) the secondary winding to ground, which will effectively reduce this noise coupling (40dB or more).
At 100MHz active solutions will not help anymore, even 35pF is a low value, it is a short at those frequencies. Anyway the coupling mechanism at those frequencies is different from the one we discuss now. The anodes of tubes start to become effective mono-pole antenna's, just to give an example.
best,
Hi Rod, I get lost! I would like to use your circuit for the filament of a single plate 2A3 . Witch circuit I should use? Can you please post the schema again in the case? Thank you
Regards
Francesco
Regards
Francesco
Contrary opinion
Question: WHY?
Everybody here who has played around with AC filament heating claims that it is sonically the best filament heating method BY A WIDE MARGIN! In AC DHT heating the impedance is very low (only secondary wire resistance), if your assumption was correct, then AC heating would give the worse sound, something that is not reality. Correct? Or at least something would be bad there. But reality is totally different! So your assumption should have something wrong...
TOTALLY WRONG!!!
For 5V 2A filaments, the filament resistance is 2.5 Ohms, right?
0.1uF cap gives 159 Ohms of impedance in 10 kHz, paralleled to the above-mentioned 2.5 Ohms. Your assumption is wrong, as only 1,5 percent of current is drawn through the 0.1uF cap, and this happens at 10kHz! At 1kHz, this is 0,15%!
Current source has the only advantage of keeping the impedance between + and - poles fairly high, so less current is passed through, at the expense of complicated electronics, which have definately a bad influence in sound quality of a DHT.
Just my 2 cents!
Question: WHY?
Everybody here who has played around with AC filament heating claims that it is sonically the best filament heating method BY A WIDE MARGIN! In AC DHT heating the impedance is very low (only secondary wire resistance), if your assumption was correct, then AC heating would give the worse sound, something that is not reality. Correct? Or at least something would be bad there. But reality is totally different! So your assumption should have something wrong...
TOTALLY WRONG!!!
For 5V 2A filaments, the filament resistance is 2.5 Ohms, right?
0.1uF cap gives 159 Ohms of impedance in 10 kHz, paralleled to the above-mentioned 2.5 Ohms. Your assumption is wrong, as only 1,5 percent of current is drawn through the 0.1uF cap, and this happens at 10kHz! At 1kHz, this is 0,15%!
Current source has the only advantage of keeping the impedance between + and - poles fairly high, so less current is passed through, at the expense of complicated electronics, which have definately a bad influence in sound quality of a DHT.
Just my 2 cents!
I forgot to mention that under any circumstances, I would like to evaluate your solution, as it could be a good solution to proper heating DHTs.
I am at the final stage of completion of a power amplifier. 2 stages, first is Emission Labs 30A DHT and second is RCA 801A triode, Class A2 wattage up to 5.5-6 W/ch. I have finished winding my Interstage and output transformers & power supplies. I have applied AC heating to 801A (final stage) and currently use LM317 as voltage regulator.
More interesting is the heating of 801A, I would like to compare AC heating against CCS heating based on your solution.
I am at the final stage of completion of a power amplifier. 2 stages, first is Emission Labs 30A DHT and second is RCA 801A triode, Class A2 wattage up to 5.5-6 W/ch. I have finished winding my Interstage and output transformers & power supplies. I have applied AC heating to 801A (final stage) and currently use LM317 as voltage regulator.
More interesting is the heating of 801A, I would like to compare AC heating against CCS heating based on your solution.
Hi Rod.
I'm hendra.
I've used ur heater with my PTT2S DHT preamp.
The noise n hum level better than my old one with LM317 V Reg + LM317 CCS.
Thanks for ur solution 🙂
My PTT2S now sounds more focus, staging is better, my PTT2S now more DHT thaan before 🙂
Thanks a lot
Regards,
Hendra
I'm hendra.
I've used ur heater with my PTT2S DHT preamp.
The noise n hum level better than my old one with LM317 V Reg + LM317 CCS.
Thanks for ur solution 🙂
My PTT2S now sounds more focus, staging is better, my PTT2S now more DHT thaan before 🙂
Thanks a lot
Regards,
Hendra
Hello Mr. Coleman,
I couldnt find the schematics for this design, only views of the PCB.
Could please you post (or repost) the schematics?
I couldnt find the schematics for this design, only views of the PCB.
Could please you post (or repost) the schematics?
The PCB + parts are available for DIYers to buy as a kit.
The design is based on some of the schematics I posted on this thread, I think about 1 year ago I posted some.
BUT, to make it reliable and easy to make, and low cost, and characterised for many different tubes, cost me probably hundreds of hours of development work! The schematic is not very different from the ones I published, but please understand that (while the kit is for sale) I must keep the latest schematic to myself.
Actually, it would be hard to make a few boards and buy the components, without spending more than the kit price!
Attachments
Herbie, just click on my (user) name to send me some email, or PM.
If you include an email address, I can send you the PDF manuals & Application Notes - these show how to use the regulators, and which DHTs have been tested, and price vs quantity, usw.
The picture in the last post is the new rev 3. Same circuit, a little easier to build. A lot of DIY builders are enjoying these, all around the world.