FUD, perhaps, sometimes for the sake of banter. Below are some observations, im done with you for today.
I sold R1 heater regulator modules, and they had a colossal error in them, because I 20/80 most of what i do, I went back and contacted everyone i could remember selling these boards, and shipped them new assembled units with the 337 operating as it should. I drew the board at 4AM because i made a bet with a friend from around the great lakes. i could sleep with two cans of red-bull.
Apparently i arrived at the same circuit as Guido did by just tinkering around in Multisim. And all my regulators are based on that core circuit and early versions had the same errors as Tentlabs regulators had, bad time constants for the low pass filter causing overshoot on warming tubes. I have a friend who very stubborn that is playing with Elrog 300B's and those have the characteristic flash during startup because the time constant capacitor/Low pass filter is too large.
A FET current source is nothing but a class A amplifier, with Source resistance comprising of the current sense plus 1/transconductance and a anode load comprising of the filaments Ohmic resistance plus some minor wiring parasitics that can cause instability if they are fed back through the miller capacitance of the device. On a tentlabs module the current source suppresses mains fundamentals in the same way as your circuit does (by swamping the resistance of the filament versus the impedance of the current source, causing noise attenuation. Just whacking in the lowest value of shunt resistor you can get away with here means your noise performance will suffer and you can improve them quite a bit by increasing this value beyond what is normally assembled. This also means that tentlabs regulators will not regulate at all under 50-100mA or so depending on luck of the draw with opamp offset voltage.
But because of different design philosophies you use a capacitor multiplier to take the easy way out and corpous amounts of RC filtering upstream. With those kind of measures its guaranteed that you get low noise, yes. But it speaks of a lack of imagination and active circuit design skills. I used to repair electronics during my studies, and some of the stuff i worked on was precision current sources, for electron microscopes (magnetic lenses) if those had the same noise figure as your product the electron beam would end up in another county.
Im wondering if the extra 9dB of noise performance is worth it, for outputs like 300B? Hell no, for something like a high-gain Ba triode, absolutely because of the gain structure of an amplifier.
The entire point of guido's circuit is that it can be dropped into an existing amplifier, with very few changes provided the transformer can deliver the extra current. And for that you cannot have a sense resistor that is too large or it will hurt your dropout performance. Ergo the opamp only sees a very small error signal and that causes the CCS impedance to roll of rather fast, now you can feed forward some of the noise, or use another opamp in open loop mode to jack up the error signal but this opens up its own can of worms.
I heard through the grapevine, that there where quite a few instances of the transistor shorting out, on both your module and Guidos, that led me to think of the circuit i posted before, which uses a Linear pre-regulator which will stabilize the voltage over the current source transistor. And provided its thermally coupled to the current source transistor provides some added protection.
I sold R1 heater regulator modules, and they had a colossal error in them, because I 20/80 most of what i do, I went back and contacted everyone i could remember selling these boards, and shipped them new assembled units with the 337 operating as it should. I drew the board at 4AM because i made a bet with a friend from around the great lakes. i could sleep with two cans of red-bull.
Apparently i arrived at the same circuit as Guido did by just tinkering around in Multisim. And all my regulators are based on that core circuit and early versions had the same errors as Tentlabs regulators had, bad time constants for the low pass filter causing overshoot on warming tubes. I have a friend who very stubborn that is playing with Elrog 300B's and those have the characteristic flash during startup because the time constant capacitor/Low pass filter is too large.
A FET current source is nothing but a class A amplifier, with Source resistance comprising of the current sense plus 1/transconductance and a anode load comprising of the filaments Ohmic resistance plus some minor wiring parasitics that can cause instability if they are fed back through the miller capacitance of the device. On a tentlabs module the current source suppresses mains fundamentals in the same way as your circuit does (by swamping the resistance of the filament versus the impedance of the current source, causing noise attenuation. Just whacking in the lowest value of shunt resistor you can get away with here means your noise performance will suffer and you can improve them quite a bit by increasing this value beyond what is normally assembled. This also means that tentlabs regulators will not regulate at all under 50-100mA or so depending on luck of the draw with opamp offset voltage.
But because of different design philosophies you use a capacitor multiplier to take the easy way out and corpous amounts of RC filtering upstream. With those kind of measures its guaranteed that you get low noise, yes. But it speaks of a lack of imagination and active circuit design skills. I used to repair electronics during my studies, and some of the stuff i worked on was precision current sources, for electron microscopes (magnetic lenses) if those had the same noise figure as your product the electron beam would end up in another county.
Im wondering if the extra 9dB of noise performance is worth it, for outputs like 300B? Hell no, for something like a high-gain Ba triode, absolutely because of the gain structure of an amplifier.
The entire point of guido's circuit is that it can be dropped into an existing amplifier, with very few changes provided the transformer can deliver the extra current. And for that you cannot have a sense resistor that is too large or it will hurt your dropout performance. Ergo the opamp only sees a very small error signal and that causes the CCS impedance to roll of rather fast, now you can feed forward some of the noise, or use another opamp in open loop mode to jack up the error signal but this opens up its own can of worms.
I heard through the grapevine, that there where quite a few instances of the transistor shorting out, on both your module and Guidos, that led me to think of the circuit i posted before, which uses a Linear pre-regulator which will stabilize the voltage over the current source transistor. And provided its thermally coupled to the current source transistor provides some added protection.
Finally it is clear what all this trolling and OT nonsense is about. You. We are just not paying enough attention to you! Misguided souls are expressing satisfaction with that Coleman and his regulators, when they should be asking about your mighty and inspired designs. No matter that they happen to look more than a little like they were copied from Guido, or that they don't work when the buyer receives them, you believe they should be the focus of every discussion on regulation. It's difficult to conceive of any other explanation for these ramblings: the thread has been washed away in a storm of Narcissism.
> I used to repair electronics during my studies, and some of the stuff i worked on was precision current sources
So first you spend your days finding production faults for someone else's design, and then move on to trying to peddle hack regulators that look and work like someone else's design. Some CV! And now you believe that this meagre experience allows you to lecture a Design professional, with a track record decades long?
My original design experience ranges from production line automation to instrumentation, and from Active satellite-finding antennas all the way to volume-produced Industrial quality ruggedised Handheld PCs. The PCs ran at thousands of units a month. This kind of quantity of complex electronic equipment is utterly unforgiving if not designed for the utmost reliability; try to imagine being responsible for holding up the work of hundreds of people because you let some flaw slip through. There is no experience to compare with this to prepare you to design things that are truly reliable and fit for purpose. It is perfectly clear that you have never done anything like this, and until you have, quit pretending to address me as a peer. You are not in the same street.
> I used to repair electronics during my studies, and some of the stuff i worked on was precision current sources
So first you spend your days finding production faults for someone else's design, and then move on to trying to peddle hack regulators that look and work like someone else's design. Some CV! And now you believe that this meagre experience allows you to lecture a Design professional, with a track record decades long?
My original design experience ranges from production line automation to instrumentation, and from Active satellite-finding antennas all the way to volume-produced Industrial quality ruggedised Handheld PCs. The PCs ran at thousands of units a month. This kind of quantity of complex electronic equipment is utterly unforgiving if not designed for the utmost reliability; try to imagine being responsible for holding up the work of hundreds of people because you let some flaw slip through. There is no experience to compare with this to prepare you to design things that are truly reliable and fit for purpose. It is perfectly clear that you have never done anything like this, and until you have, quit pretending to address me as a peer. You are not in the same street.
Decades of experience or not, everyone needs to keep his feet on the ground, and their ego's in check. Myself included. And the irony in your post wasn't lost on me or the reader.
What really grinds my gears is invoking seniority, authority instead of reason proof or verifiable facts. You post contradictory specifications that simply dont add up, No measurements or methods to speak of, you throw numbers around that are hard to believe. And you stonewall any comment suggestion however helpfully intended.
Hell i told you six months ago that there may be an issue with cathode poisoning using current sources, yet you give no indication of having done anything to your original product to address this issue.
Im really curious what V9 looks like now, wonder if the transistors are still flapping in the breeze, The pads on the PCB too small for comfort and no mounting holes to mount it conveniently or screw terminals that make it look like an aliexpress kit.
If you bias regulator forces the tubes into cutoff during startup thats a good thing, just dont forget to mention that a choke input supply is going to overvolt the electrolytics if solid state or a directly heated rectifier is used. Cause omission is a lie in disguise.
What really grinds my gears is invoking seniority, authority instead of reason proof or verifiable facts. You post contradictory specifications that simply dont add up, No measurements or methods to speak of, you throw numbers around that are hard to believe. And you stonewall any comment suggestion however helpfully intended.
Hell i told you six months ago that there may be an issue with cathode poisoning using current sources, yet you give no indication of having done anything to your original product to address this issue.
Im really curious what V9 looks like now, wonder if the transistors are still flapping in the breeze, The pads on the PCB too small for comfort and no mounting holes to mount it conveniently or screw terminals that make it look like an aliexpress kit.
If you bias regulator forces the tubes into cutoff during startup thats a good thing, just dont forget to mention that a choke input supply is going to overvolt the electrolytics if solid state or a directly heated rectifier is used. Cause omission is a lie in disguise.
There is nothing especially difficult about designing a DHT for 40000 hr operation. Plenty of getter, and high quality materials. Modern production often falls far short of this, due to cost-cutting in the materials and slapdash assembly methods.
Specifying lifetime without also specifying power ON-OFF cycles is meaningless to practical usage though.
You can build a copy of 1930s WE amplifiers, is you like. But I don't really accept that there would be any authenticity in the sound of this, unless you confine yourself to playing 78rpm shellac records.
Specifying lifetime without also specifying power ON-OFF cycles is meaningless to practical usage though.
You can build a copy of 1930s WE amplifiers, is you like. But I don't really accept that there would be any authenticity in the sound of this, unless you confine yourself to playing 78rpm shellac records.
Of course that begs the question: why are we so enamoured of the DHT in the first place, especially the 300B? Is it for a measure of authenticity, or are we hearing something that is not available in the indirectly heated triodes?
Last edited:
The large-signal linearity of the best DHTs surpasses any other device, open loop. Draw a load line on different sets of tubes curves to see for yourself. Open the lines into an ellipse, to accommodate the reactive loads of a loudspeaker.
Or simply measure your work, at the output terminals.
Or simply measure your work, at the output terminals.
I've been on the verge of purchasing a pair of matched WE 300Bs from the Rossville works for a month or so. I don't know if they are worth the money, which will include Customs duty, VAT, shipping, and Revenue handling charges. I just can't make up my mind (it's a big outlay). I'll probably use your filament regulator in any case.
Mentioning expensive 300Bs, I find the JJs sound dry and wooden. I have a selection of Chinese 300Bs, and my pair of matched Golden Dragons sound good, and way better than the JJs.
What do you think about WE 300Bs - have you heard them?
Last edited:
IMHO the difference in sound between 300B types is smaller than difference between good and EXCELLENT OTs, anode supply, Filament supply, and driver stage design and power supply.
So if you don't have a Sowter, or Monolithic OT, I would start there. OTs - even high cost ones are good value, and last a lifetime.
So if you don't have a Sowter, or Monolithic OT, I would start there. OTs - even high cost ones are good value, and last a lifetime.
Filament Bias is a simple technique to apply cathode bias, but with a low-value resistor. The goal for doing it is to have a low enough value of cathode resistor, that the bypass capacitor is no longer needed.
For valves with reasonably low filament current, and low bias voltages, it is a good solution. But for something like a 2A3 (45V bias, 2.5A current: = >110W burned in the resistor) the heating power in the filament bias resistor is far too much.
Before building any stage with filament bias, be sure to check the power burned in that resistor. If the power is too high, grid bias, aka 'fixed bias' is the best solution that allows the bypass capacitor to discarded.
IMPLEMENTATION: My application note for Filament Bias - giving instructions on how to build it for the 26 or 4P1L, including transformer choices and Raw DC design, and a sample preamp circuit are given here:
http://lyrima.co.uk/pdf/FilamentBias_Supply.pdf
For valves with reasonably low filament current, and low bias voltages, it is a good solution. But for something like a 2A3 (45V bias, 2.5A current: = >110W burned in the resistor) the heating power in the filament bias resistor is far too much.
Before building any stage with filament bias, be sure to check the power burned in that resistor. If the power is too high, grid bias, aka 'fixed bias' is the best solution that allows the bypass capacitor to discarded.
IMPLEMENTATION: My application note for Filament Bias - giving instructions on how to build it for the 26 or 4P1L, including transformer choices and Raw DC design, and a sample preamp circuit are given here:
http://lyrima.co.uk/pdf/FilamentBias_Supply.pdf