The quality and durability of the Emission Labs EML range of DHTs is well worth highlighting.
They are a high price; but the construction quality gives confidence. My EML300bs have run seven years with little degradation, EHX equivalents sometimes failed completely in the same number of months.
You can drive the huge output DHTs at the >200Vpp they need, with low distortion.
Kevin mentioned the EML20B; the 20 & 30B series are difficult to match for very high output voltage swings.
The curves tell the tale: & measured spectrum matches it (Vegard posted here )
I give them all possible credit for the great curves, but at such a surprisingly low perveance. A modern high perveance valve, like a triode-connected hot pants video pentode, something in the 12HL7 family or the 6Zh9pi or 6Zh11pi or 6JC6 family gives great curves at an easier geography on the plate curves. At the same region of the plate curves, 6SN7s are pretty linear too. Actually, pretty similar.
So there must be some extra factor, or folk wouldn't make the extra effort. But what?
Much thanks, as always,
Chris
So there must be some extra factor, or folk wouldn't make the extra effort. But what?
Much thanks, as always,
Chris
You'd have to be more specific.
In keeping with this thread: driving say, 845, with >200Vpp (some drive them with non-negligibly more) the EML20B is very tough to match. Just look at the EML20B Curves, and impose this swing on them - whichever kind of load you choose. The output spectrum is equally marvellous.
Another important point: The (operating point) grid voltage of the EML20B is -10V or even greater: allowing direct connexion to high-output DACs: with high performance units now going way above 0dB = 2.0V rms, this is not a trivial consideration.
Try the same output swing on the 6SN7 Curves. Not failing to note that the 6SN7 anode ra slope is not even constant across the swing either. I found that 6SN7s pressed into driver service sound muddled, and lack articulation, as an aside.
In keeping with this thread: driving say, 845, with >200Vpp (some drive them with non-negligibly more) the EML20B is very tough to match. Just look at the EML20B Curves, and impose this swing on them - whichever kind of load you choose. The output spectrum is equally marvellous.
Another important point: The (operating point) grid voltage of the EML20B is -10V or even greater: allowing direct connexion to high-output DACs: with high performance units now going way above 0dB = 2.0V rms, this is not a trivial consideration.
Try the same output swing on the 6SN7 Curves. Not failing to note that the 6SN7 anode ra slope is not even constant across the swing either. I found that 6SN7s pressed into driver service sound muddled, and lack articulation, as an aside.
I will say the EML20B is the most linear DHT I have worked with, in my line stage at a couple of volts RMS out 20K:600 (5.8:1) amorphous core output transformer I have measured THD in the vicinity of 0.007% 100Hz - 10kHz without feedback (using my RTX) and that is pretty impressive.
In 2021 banpuku's source was: dac - 2 Volt output - passive Emmia preamp (150H autoformer from Dave Slaggle)
Would you connect that autoformer to a 2 stage SE EML20B - GM70 (520B) amp, if the amp has a step-up input-transformer ?
Would you connect that autoformer to a 2 stage SE EML20B - GM70 (520B) amp, if the amp has a step-up input-transformer ?
That is a question for Dave, I am somewhat familiar with his autoformers - performance would depend very heavily on the source impedance of the autoformer on the taps most commonly used and the effective input impedance of that transformer. It could work but might not be optimal.
Re: the EMLs : You guys are probably (well, maybe somewhat more than probably) right about the EML drivers. If you have 1kV already available, why not take the best possible advantage of it? If the driving valve can operate safely and linearly off the right side of the curves, why not? My prejudice to a certain mindset (high perveance, big cathode) is not the only way to carve the turkey.
Can't really quite get to the mythical single gain stage between 100K volume control and 1kV valve's grid with a mu of 12 family (211, 304TL) anyway, let alone mu of 5 (845, GM70?, etc.). With a mu of 50 or 60 (G1-G2 triode) video pentode it can be done, but not with as nice curves as the EML, but it, with a mu of 20 still needs an extra 10 to 20 dB gain, so another stage. Horses for courses and ho, ho, ho.
All good fortune,
Chris
Can't really quite get to the mythical single gain stage between 100K volume control and 1kV valve's grid with a mu of 12 family (211, 304TL) anyway, let alone mu of 5 (845, GM70?, etc.). With a mu of 50 or 60 (G1-G2 triode) video pentode it can be done, but not with as nice curves as the EML, but it, with a mu of 20 still needs an extra 10 to 20 dB gain, so another stage. Horses for courses and ho, ho, ho.
All good fortune,
Chris
If we are designing a GM-70 or 845 SE amplifier with EML20B driver, we're likely to be going all-out for performance (sound), and unlikely to be put off by parts cost, or apparent difficulty. A single-stage driver is perfectly possible.
With the output stage biased at around -100V, and driver µ of 20, the 0dB input must be at or above 5.0V (zero-to-peak). that's only a little more than the output of the canonical CD player (2V rms, ~2.8V zero-to-peak).
So the goal should be to avoid a third stage. I think there are a few better ways to get the right input to the EML20B:
Simply using a high quality, high-output DAC, with DSP volume control, the output is often high enough already.
Or, a 2:1 SE IT can be used on the input.
Or the Post-DAC reconstruction filter can be examined and updated. It's not unusual for the differential stage to be attenuating to allow the first stage to achieve better noise performance through higher gain (see for example the TI App note on this here, see design step 3 ) so a better-tailored version of the second stage could be designed for higher output.
Or, the op-amp differential stage of the DAC could be replaced with a 1+1:1+1 differential to SE-connected transformer, to great effect. Ater all, the source impedance is very low. The IT does not care much about any broadband noise that escapes the first stage filters and ITs are more in keeping with the EML20B driver. (The differential stage must always be retained as differential, to cancel the common-mode distortions from the DAC).
With LP systems, increasing the gain of the RIAA stage is a better use of your high-gm tetrodes; but in many cases a DHT line stage is already present, with 10dB or more gain, to drive the cables properly.
With the output stage biased at around -100V, and driver µ of 20, the 0dB input must be at or above 5.0V (zero-to-peak). that's only a little more than the output of the canonical CD player (2V rms, ~2.8V zero-to-peak).
So the goal should be to avoid a third stage. I think there are a few better ways to get the right input to the EML20B:
Simply using a high quality, high-output DAC, with DSP volume control, the output is often high enough already.
Or, a 2:1 SE IT can be used on the input.
Or the Post-DAC reconstruction filter can be examined and updated. It's not unusual for the differential stage to be attenuating to allow the first stage to achieve better noise performance through higher gain (see for example the TI App note on this here, see design step 3 ) so a better-tailored version of the second stage could be designed for higher output.
Or, the op-amp differential stage of the DAC could be replaced with a 1+1:1+1 differential to SE-connected transformer, to great effect. Ater all, the source impedance is very low. The IT does not care much about any broadband noise that escapes the first stage filters and ITs are more in keeping with the EML20B driver. (The differential stage must always be retained as differential, to cancel the common-mode distortions from the DAC).
With LP systems, increasing the gain of the RIAA stage is a better use of your high-gm tetrodes; but in many cases a DHT line stage is already present, with 10dB or more gain, to drive the cables properly.
Thanks Rod. Do you have any comments about the reason that these "old-fashioned" and certainly fiddly DHT triodes have such nice curves? Is it related to the lower perveance, maybe related to the grid-cathode relationship (or something else?) or do the nice curves come from design choices that maximize linearity at difficult (but not impossible) regions of the plate curves?
The next size up, "50 watters" like type 211 and 845 may give some clue to the answer. We (well, me) easily accept their voltage and fiddly filament demands, for the sake of performance. Maybe I need to extend this acceptance to the driver stage. The penalties in the lack of a separate cathode might be worth the pain.
But the question remains for me, why do these old-fashioned buzzards have such nice curves? Can't be voodoo, must be something interesting.
All good fortune,
Chris
ps: everything about changing the sources, etc. is cheating. So there. Gotta be: 100K Ohm volume control, no stray capacitance, to big-butt triode 100-ish VRMS, single stage. Ready, set, go!
pps: why do I get to make the rules? because it's my football.
The next size up, "50 watters" like type 211 and 845 may give some clue to the answer. We (well, me) easily accept their voltage and fiddly filament demands, for the sake of performance. Maybe I need to extend this acceptance to the driver stage. The penalties in the lack of a separate cathode might be worth the pain.
But the question remains for me, why do these old-fashioned buzzards have such nice curves? Can't be voodoo, must be something interesting.
All good fortune,
Chris
ps: everything about changing the sources, etc. is cheating. So there. Gotta be: 100K Ohm volume control, no stray capacitance, to big-butt triode 100-ish VRMS, single stage. Ready, set, go!
pps: why do I get to make the rules? because it's my football.
banpuku - sorry if i am impolite now:
You have got good advice from Rod Coleman, KevinKr, andyjevans, euro21 .... In your threads of the last 2 years you have got dozens of good technical advice from people with very good technical knowledge. You have ignored most of it.
A few examples:
You knowledge of tube basics is so low after 15+ years of tube diy, that it is dangerous. How could we expect, that you will understand our suggestions. Once you have reported, you was electrocuted at tube diy, survived and needed long time to recover.
You are tough, don't give up and try it again;
Fine, but once your reserves of good luck in life will be gone .....
You want "absolutely" to build your own all direct heated triodes single ended power amp. But you are not able to design it.
Here on the forum are a lot of talented people.
Ask them for help - to design you a complete amp schematic, psu and how to wire it.
Maybe one is willing and has time to do it for you.
Don't expect to get it for free - for a premium schematic you have to pay ....
I am sure not one of that talented people;
But little as my knowledge is, it is good that i know where my limits are.
You have got good advice from Rod Coleman, KevinKr, andyjevans, euro21 .... In your threads of the last 2 years you have got dozens of good technical advice from people with very good technical knowledge. You have ignored most of it.
A few examples:
- They suggested you can not to load the Aa tube rp=30K with a 5K interstage, you did it.
- You have asked for help, because there was no AC signal passing the coupling cap from the Aa to the power tube. Well you had connected the Aa's 70K loading resistor to the ground instead of B+ voltage.
- In your "low dcr choke" thread you wanted to change the selfbias EML45B to fixed bias. You had a fixed bias kit from Rod Coleman with pcbs, schematic and tips how to wire the negative psu. Even with the help of this forum you could not do it.
You knowledge of tube basics is so low after 15+ years of tube diy, that it is dangerous. How could we expect, that you will understand our suggestions. Once you have reported, you was electrocuted at tube diy, survived and needed long time to recover.
You are tough, don't give up and try it again;
Fine, but once your reserves of good luck in life will be gone .....
You want "absolutely" to build your own all direct heated triodes single ended power amp. But you are not able to design it.
Here on the forum are a lot of talented people.
Ask them for help - to design you a complete amp schematic, psu and how to wire it.
Maybe one is willing and has time to do it for you.
Don't expect to get it for free - for a premium schematic you have to pay ....
I am sure not one of that talented people;
But little as my knowledge is, it is good that i know where my limits are.
Definitely no voodoo.
We were admiring the EML20B: and might remark that this is a rare thing - new-design of DHT, offering something not previously availble. Luckily, the designer had soemthing to say about the design aim, and how it was realised.
They say: (source link here)
So yes, the need for high operating voltages is also part of the deal, to achieve low distortion.
It is the high anode voltage region of the anode curves that divide DHTs from the unipotential (indirectly heated) alternatives, and demonstrate the better performance. The move to unipotential cathodes is a form of 'miniaturisation' in itself, since the cathode is compressed into a thin cylinder, where the DHT has a wide area distributed cathode. This choice is forced on the indirectly heated designs, since the heating power is already larger, and widening the cathode area would likely drive it to excessive levels.
IF you insist. To satisfy this route, simply use the EML30A version with its µ of 32.
But I think that using a DAC with differential outputs (including available commercial units, or, modest-difficulty DIY) and connecting the IT to it (1+1:1+1SE) will yield multiple valuable advantages, and is to be preferred....
Hi everyone, thanks for the continued conversation. To help answer a few questions about the project:
1. DC filament heaters are sourced by Rod's excellent v9 boards. Getting great results with < 2mV noise measured at the speaker terminals. With my 90db speakers, this is completely inaudible.
2. Agree that GM70 supply is becoming more scarce, especially for Copper plates.
3. my existing source is a DAC (output 2.0V) --> Emia Autoformer --> monoblocks
4. these monoblocks are up and running. Sound great (best I have heard of any product (DIY or Commercial) in my system).
5. The original post regarding current calculations was to determine if the Aa had the drive current for the GM70s. Based upon input on this thread, it appears not. That is OK, as I can change out the driver.
6. The interstage transformer is a custom made Monolith Magnetics 200H 1:1 nano crystalline IT. Huge beast. Beautiful sound.
Now, what to do about the Aa? I have tried an 841 and C3g as drivers for the GM70. The sound of the Aa is better in my opinion. The Aa is much more nuanced, open, airy and has unparalleled tonality. Are there better, more appropriate drivers than the Aa? Possibly. Let's continue the hunt!
1. DC filament heaters are sourced by Rod's excellent v9 boards. Getting great results with < 2mV noise measured at the speaker terminals. With my 90db speakers, this is completely inaudible.
2. Agree that GM70 supply is becoming more scarce, especially for Copper plates.
3. my existing source is a DAC (output 2.0V) --> Emia Autoformer --> monoblocks
4. these monoblocks are up and running. Sound great (best I have heard of any product (DIY or Commercial) in my system).
5. The original post regarding current calculations was to determine if the Aa had the drive current for the GM70s. Based upon input on this thread, it appears not. That is OK, as I can change out the driver.
6. The interstage transformer is a custom made Monolith Magnetics 200H 1:1 nano crystalline IT. Huge beast. Beautiful sound.
Now, what to do about the Aa? I have tried an 841 and C3g as drivers for the GM70. The sound of the Aa is better in my opinion. The Aa is much more nuanced, open, airy and has unparalleled tonality. Are there better, more appropriate drivers than the Aa? Possibly. Let's continue the hunt!
Rod & Kevin, have you heard the EML30A as a driver in a system? curious about your subjective thoughts about its sonic virtues and shortcomings. Thanks, Pat
I don't use a 30A driver, but I have compared the sound of EML300B to lots of other 300Bs in the same socket. The EML has constructional improvements and higher quality than all of them. I also have new WEs, but have not yet burned them in or compared them. The EML is so far my favourite: it is neutral and revealing, and brings great life to any music.
THe EML is also the only 300B that immediately stops the music when you lift the stylus from the record. The others have a resonant overhang of sound. The EML has extra bracing of the Anode, among its structural improvements
There are others here on diyAudio who have used the 20/30A series, hopefully someone will add something about the driver.
But in any case you have a unique environment, and other opinions may not carry across to your system.
But we can see if the 30A will actually work in your system:
Firstly, is the input autoformer step up down, of both? You need step down for the Aa, or it might clip the input, at the data sheet op point Ua = 220V, Ug = -2V, Ia = 3mA. 2.0V rms input means: over 2.8V zero-to-peak.
As for the EML30A, you could choose the data-sheet point of 420V, -6.7V, 10mA, ra = 8.5kΩ; µ = 36.
Without any input IT, a 2.0Vrms DAC at 0dB will give 102V zero-to-peak. IF your GM70s are biased to -102V or less, it will work, using the DSP volume, without the IT. Or, you can use the IT to give something in hand, for a volume control.
GOing from the ra of 30k [for Aa] to the 30A's 8.5k (or lower, for hotter op points) will certainly be an improvement, performance-wise, and the large anode of the 30A will show better curves at the high voltages.
These are things that usually give improvements to the sound too; but only trying it can show.
THe EML is also the only 300B that immediately stops the music when you lift the stylus from the record. The others have a resonant overhang of sound. The EML has extra bracing of the Anode, among its structural improvements
There are others here on diyAudio who have used the 20/30A series, hopefully someone will add something about the driver.
But in any case you have a unique environment, and other opinions may not carry across to your system.
But we can see if the 30A will actually work in your system:
Firstly, is the input autoformer step up down, of both? You need step down for the Aa, or it might clip the input, at the data sheet op point Ua = 220V, Ug = -2V, Ia = 3mA. 2.0V rms input means: over 2.8V zero-to-peak.
As for the EML30A, you could choose the data-sheet point of 420V, -6.7V, 10mA, ra = 8.5kΩ; µ = 36.
Without any input IT, a 2.0Vrms DAC at 0dB will give 102V zero-to-peak. IF your GM70s are biased to -102V or less, it will work, using the DSP volume, without the IT. Or, you can use the IT to give something in hand, for a volume control.
GOing from the ra of 30k [for Aa] to the 30A's 8.5k (or lower, for hotter op points) will certainly be an improvement, performance-wise, and the large anode of the 30A will show better curves at the high voltages.
These are things that usually give improvements to the sound too; but only trying it can show.
banpuku
In your "50/50 SET" thread from April 2020, you have got a good advice (from vega65) to use the EML 30A as driver with a IT. Why did he suggest you a medium rp + high current +32 gain DHT tube as driver ? There are no "NOS DHT" with all these 3 characteristics.
With NOS DHT you can build a 3 stage power amp, but not 2 stage as you want.
Your Monolith nano crystalline 1:1 IT with 200H = ca. 25K/25K . A good load for a EML 30A. (Not for the rp 30K Aa tube)
What DC current can the IT handle ? Not max DC, bias DC. A bias of ca. 20mA of the 30A would drive the GM70 better.
Soundwise, no experience with EML 30A.
In your "50/50 SET" thread from April 2020, you have got a good advice (from vega65) to use the EML 30A as driver with a IT. Why did he suggest you a medium rp + high current +32 gain DHT tube as driver ? There are no "NOS DHT" with all these 3 characteristics.
With NOS DHT you can build a 3 stage power amp, but not 2 stage as you want.
Your Monolith nano crystalline 1:1 IT with 200H = ca. 25K/25K . A good load for a EML 30A. (Not for the rp 30K Aa tube)
What DC current can the IT handle ? Not max DC, bias DC. A bias of ca. 20mA of the 30A would drive the GM70 better.
Soundwise, no experience with EML 30A.
Good feedback, Lauscher. Thanks. The IT can handle 8mA which 2x more than the Aa. The EML 30A can be biased at 300V, -4V and 8mA. Not sure if it's worth the 8mA vs. 4mA, or do I look elsewhere? The other option that I am considering is using a 10Y with a 1:4 input transformer.
You could try the 10y, if you have a 1:4, but that feels like a somewhat marginal solution. I wouldn't recommend spending big on the IT.
The 30A can run 8mA, but I suggest higher anode voltage, so that the idle grid voltage is big enough to accommodate the DAC output voltage. 420V anode, I think.
Vegard (vega65) builds good things; if he recommended the 30A I'd take that seriously.
The 30A can run 8mA, but I suggest higher anode voltage, so that the idle grid voltage is big enough to accommodate the DAC output voltage. 420V anode, I think.
Vegard (vega65) builds good things; if he recommended the 30A I'd take that seriously.
If the Monolith IT can only handle 8mA, it makes no sense to use a EML 30A to drive the GM70. Look at all the threads here, on how to drive a big transmitting tube - they all try to use a low rp + medium to high current driver.
Your amp, Aa loaded with the 25K Monolith IT, i assume it lacks bass.
You would need a very low output impedance dac + strong AC-signal current to drive a step-up input-transformer in your power amp. Sorry, i am no native english speaker, it is to complicate for me to explain what kind of nasty things a input step-up can do to audio signal, if not driven from proper source.
This kind of question you have already asked a few times in the past. You might use a high voltage output dac.
Look how often you have bought the wrong transformers, tubes - all very expensive. (and to sell it)
I can't help.
Your amp, Aa loaded with the 25K Monolith IT, i assume it lacks bass.
You would need a very low output impedance dac + strong AC-signal current to drive a step-up input-transformer in your power amp. Sorry, i am no native english speaker, it is to complicate for me to explain what kind of nasty things a input step-up can do to audio signal, if not driven from proper source.
This kind of question you have already asked a few times in the past. You might use a high voltage output dac.
Look how often you have bought the wrong transformers, tubes - all very expensive. (and to sell it)
I can't help.
Lausher - I get your point. Very helpful. Regarding bass: my room is full of bass, BUT the tightness / quality of the bass is not first class once we get below roughly 60-80hz. Then, the bass becomes less defined and loose. Otherwise, I am very happy with the upper bass and lower midrange.
Here is my general question and why I started this thread. I understand that "technically" the Aa is a poor driver choice for the GM70. But, it sounds damn good. To my ears, best I have heard in my system. BUT, in the back of my mind there is this little voice saying "you might be missing something by using the Aa driver". So, I am trying to understand what "sonic" compromises the low current driver would cause for the GM70. I guess I should have asked this in the OP.
Here is my general question and why I started this thread. I understand that "technically" the Aa is a poor driver choice for the GM70. But, it sounds damn good. To my ears, best I have heard in my system. BUT, in the back of my mind there is this little voice saying "you might be missing something by using the Aa driver". So, I am trying to understand what "sonic" compromises the low current driver would cause for the GM70. I guess I should have asked this in the OP.