I have exactly one lonely 306A looking for a mate. Just can't find one anywhere for a couple of years...
Rod, I agree 100%! It'd be nice to see some real measured curves. The attached are from Pete Millett's site, the only sample I could find.
Attachments
I agree that GM-70 type tubes and headphones seem a risky proposition. I believe I checked with Tribute and they do not do custom windings, so asking for an 8+8+8+8 secondary would be impossible, correct me if I am wrong.
PMillet and Eddie Current both designed headphone amps with the 307A that were well received. The Eddie current model even had the ability to switch to 300B's and several reported liking the 307A's better. I know that's just anectodal evidence, but a thread here reported that the screen is fine operating at 20W's (I think Tubelab reported 25W.) I have nothing against Sovetek 300B's or any other 300B but I haven't seen any independant tube tracing of these either, so I'm going with the $12 tubes. I agree though that the design should accomodate both just in case.
Yes, Let's hope for a 307-A and a Tracer meeting up soon.
The other route, if a bit more headroom is the goal, is via the EML super 300Bs: 300B-XLS, EML 520 and 1605.
Emission Labs - Introduction
you can pick your operating point from 360V/60mA to 600V or more, and all points between.
This way, we can avoid the high [800 ... 1000V] voltage of the GM-70, and its high-Z transformer preference. The big EMLs can work with 4500 ohm or less.
I know they are costly, but given the cost of the headphones, and the transformers, going after bargain tubes seems out-of-line with the rest of the project, IMHO.
For example, one of the moderate operating points for the EML 300BXLS is:
475V - 90mA - 4200 ohm anode load........... for 14W ouput.
We would be unlucky not to be able to deliver 5W cleanly from this, even with terribly inefficient transformer solutions (though I would solve this problem by using the Sowter custom Transformer - the offer mentioned at the start of the thread, or the Tribute Custom OT).
This platform can be used easily with any old 300B to get started. The 4200ohm load is near enough to optimum, and the voltage is perfect when used with cathode bias (etc) giving about 395V A-K...perfect! My Filament heating kits can be supplied to switch between 2A and normal 300B [1.2 .. 1.3A] filaments, easily enough.
This way, the whole design can be built with a ready upgrade route, for those who prefer incremental builds, or just a bit more choice.
475V - 90mA - 4200 ohm anode load........... for 14W ouput.
We would be unlucky not to be able to deliver 5W cleanly from this, even with terribly inefficient transformer solutions (though I would solve this problem by using the Sowter custom Transformer - the offer mentioned at the start of the thread, or the Tribute Custom OT).
This platform can be used easily with any old 300B to get started. The 4200ohm load is near enough to optimum, and the voltage is perfect when used with cathode bias (etc) giving about 395V A-K...perfect! My Filament heating kits can be supplied to switch between 2A and normal 300B [1.2 .. 1.3A] filaments, easily enough.
This way, the whole design can be built with a ready upgrade route, for those who prefer incremental builds, or just a bit more choice.
I only used Tribute's example because I remember Pieter citing a very low DCR value for the 8R secondary of a 300B transformer. As I am sure he interleaves his trafo, I just assumed 4 windings, which in parallel result in 0,2R, and that in series parallel would be 0,8R DCR. But that is just some assumptions.
Thanks Bas, thanks Pieter,
I was indeed not sure exactly how high the DCR figure on the secondary was, so good to have new real numbers. But is my thought right that, if one would apply the above transformer to obtain a 32R output, the DCR would increase by a factor 4?
40R DCR for 30H in a SE transformer. That is impressive! Makes me wonder how low primary DCR could be for a PP transformer of similar power rating and inductance.
Somehow I confused Tribute with another european supplier, it is good news to know there is a tranformer supplier who does custom work with high efficiency C-core. That low DCR primary is ideal, is that an amorphous core?
I do like the concept of asking for two secondaries to give balanced output. Two 8 secondaries gives us a 32 out impedance match. I think that this would give us CMRR at the voice coil? This may be beneficial as we move toward the 300B direction as the B+ for a 300B is pushing the envelope for the SSHV, but I loathe to giving up to the SSHV for this project, it gives a low output impedance that really makes SET sound terrific I've had such good results with the SSHV.
I did some research on what 300B's cost and supposedly sound better than others, typical audiophile stuff and it just left a bad taste in my mouth. Something I don't like about not being able to buy NOS, but I fully admit to being a novice in the field of expensive power triodes and prefer to try cheaper alternatives.
Just noticed that the AN HE 6 kit uses two EL34's in PSE, For some reason I thought this was a single EL34 per channel. This is really a speaker amp, probably a nice one. And there are NOS tubes, one could buy a quad of the NOS russian equivalent for the cost of one 300B. But I don't think all this is worth the trouble for most just of us for a single headphone (HE-6).
The newer model HE-500 sounds great, they are like Maggies for the head (planar magnetics.) I've been satisfied with my spud WE417 SET powering these (low volume) over a higher power P-P amp I own. Probably the HE-6 will be replaced with a more efficient HE-7 before I complete this project anyway.
This is the 307A info, "operating at 400V 45mA"http://www.diyaudio.com/forums/tubes-valves/68216-where-307a-vt225-amps-5.html#post1707179.
There isn't a doubt in my mind it is more linear than a 2A3 and probably budget modern 300B's. But...
Ale, I could ship you one NOS VT-225 if you want to trace it and share the model and send her back. It would be interesting to compare to a Sovtek 300B
I do like the concept of asking for two secondaries to give balanced output. Two 8 secondaries gives us a 32 out impedance match. I think that this would give us CMRR at the voice coil? This may be beneficial as we move toward the 300B direction as the B+ for a 300B is pushing the envelope for the SSHV, but I loathe to giving up to the SSHV for this project, it gives a low output impedance that really makes SET sound terrific I've had such good results with the SSHV.
I did some research on what 300B's cost and supposedly sound better than others, typical audiophile stuff and it just left a bad taste in my mouth. Something I don't like about not being able to buy NOS, but I fully admit to being a novice in the field of expensive power triodes and prefer to try cheaper alternatives.
Just noticed that the AN HE 6 kit uses two EL34's in PSE, For some reason I thought this was a single EL34 per channel. This is really a speaker amp, probably a nice one. And there are NOS tubes, one could buy a quad of the NOS russian equivalent for the cost of one 300B. But I don't think all this is worth the trouble for most just of us for a single headphone (HE-6).
The newer model HE-500 sounds great, they are like Maggies for the head (planar magnetics.) I've been satisfied with my spud WE417 SET powering these (low volume) over a higher power P-P amp I own. Probably the HE-6 will be replaced with a more efficient HE-7 before I complete this project anyway.
This is the 307A info, "operating at 400V 45mA"http://www.diyaudio.com/forums/tubes-valves/68216-where-307a-vt225-amps-5.html#post1707179.
There isn't a doubt in my mind it is more linear than a 2A3 and probably budget modern 300B's. But...
Ale, I could ship you one NOS VT-225 if you want to trace it and share the model and send her back. It would be interesting to compare to a Sovtek 300B
Indeed, from 8 to 32 ohm the DCR would increase by a factor of 4. However the efficiency (or power loss) of the transformer and the damping factor don't change until you use all the windings.
For instance, if you have 4 identical secondaries all in parallel to give 0.2 ohm for 8 ohm than each of them will measure 0.8 ohm. For 32 ohm the DCR will be 0.8 ohm because you need to put them two by two in parallel and then the two resulting windings in series.
Last edited:
To obtain a low dcr is very easy if you choose a big core, planty space for copper
I've always been taught that the bigger the core the poorer the SQ of small signals. With headphones having the driver right up next to the ear and most ultra high efficiency we are dealing with tenths and hundredths of millwatt sound. Its why a mic output transformer is usually a small mu-metal transformer.
The new orthos throw wrench into the equation in that we want not only low DCR but high efficiency and we need inductance without resorting to a 100mA core? With headphones we are dealing with voltage swings that are on par or higher than a Klipshorn for example but less current and much less damping required.
Wouldn't 4 secondary's in series give a higher insertion loss than a single 32 ohm secondary? There is some advantage to a single secondary, I think you get better coupling and hence better transfer of the low level mid/treble energy. Also taps cause issues I avoid them.
If a hypothetical 5k:32 transformer had a .8x4= 3.2 ohm DCR this would be very good compared the 12 ohm DCR 32 ohm secondary on my spud set but we are still powering the phone with 4 times the inductance in series. Trying to simulate an SET OPT I see odd things from secondary DCR like standing DC. Just saying there is a trade-off, and we know 8 ohm secondary's have been "perfected" over the years.
That's why I like the dual 8 ohm balanced drive idea on paper and simulations the best. But I don't know if it would take away the SET sound? Would the balanced drive reduce 2H distortion at the driver ?
Thanks for the replies from Regal, esltransformer and 45. Indeed, bigger core = less turns for same inductance (but OTOH mean lenght of turns is higher in comparison to smaller core). And to 45 thanks for confirming the DCR ratio's for a 32R tap.
All with all it seems possible to get the necessary power from a 300B keeping low secondary DCR. Still, I think I would go with PP The nice thing about the dual 8R secondary is that the amplifier can be used for both speakers and HE-6 without (serious) performance compromises on either of them, and that makes it much more attractive to invest in a good set of OPTs as you will need only one pair for all listening.
All with all it seems possible to get the necessary power from a 300B keeping low secondary DCR. Still, I think I would go with PP
The nice thing about the dual 8R secondary is that the amplifier can be used for both speakers and HE-6 without (serious) performance compromises on either of them, and that makes it much more attractive to invest in a good set of OPTs as you will need only one pair for all listening.
Dave, My enthusiasm for the 300B is for offering a design blueprint that everyone can join in with, and build from something like a tried & tested schematic.
300B is a winner here, on account of the great choice of tubes a DIYer can buy - from low-cost Shuguangs, via excellent mid-price EH and JJ, up through TJ/Fullmusic and into EML super-DHT territory.
The 300B also meets all of your design-goals:
- Linear DHT;
- can always deliver 5 clean watts;
- upgrade route through the super-300Bs for more power, or to compensate for OT losses;
- easy-obtanium (but not lo-pecuniam);
If the 307A can really survive 400V for a usable operating life, then it can probably take the same OT and power supply. Maybe it takes the place of a low-cost entry level DHT for curious experimenters, which just adds to the fun.
I know that new 300Bs cost a lot more than old-stock from Antique dealers, but I have always paid for them gladly. The JJs that I bought in 1998 worked daily for 10 years - so that's £10/year (Euro 11; USD 15/year). Pretty low-cost, as pleasure goes.
My impression of the new-model EH 300B Gold is even better - it is much better-aligned than the JJ.
As for measured curves: try comparing Audiomatica's characterisation of the Richardson (Cetron) - which was a roughly-made unloved dog -
The 300B Tube
with the new EML 300B:
EML 300B Data sheet. 22.2.08
Look very close in terms of linearity.
I believe you will find that they all measure very similarly. And provided they have correctly copied the WE internal geometry, they will. This large scale 300B geometry is far more responsible for the linear nature of the DHT, I suspect, than anything in the materials.
They all sound different, I know:
- quality of vacuum/gas. Might have given old samples an advantage, but not now, after 70 years of storage. Gas-induced grid current is noisy, the last thing you need in a headphone amplifier.
- microphony. I can vouch for JJ (older) and EH as acceptable, but it's never too good.
- Dependency on filament supply stability. EML made great fun out of the TJ/Fullmusic 300B mesh for this property. With the TJ, the anode-current goes up & down with filament voltage. This is despite its high cost, and reputation.
I developed my filament regulator on 300Bs, and one outcome was that the (old, bad quality version ) Sovtek was made to sound almost as good as the JJ, & I suspect that it had a lot to do with this effect. Filament supply must always be high quality, but with some versions of 300B, the improvements are magnified, by eliminating this problem.
Anyway, that's my submission: to meet the goals of the project, an OT with a 60mA dc current handling and 4500 to 5000 ohm primary. That way, anyone can join in, and use the same transformer. It should work with 307A, should they prove equal to the task, and will certainly work with standard 300B operating points for 320V to 400V. If the need arises for more power, just increase B+ to 450V, 500V or even 600V and use the appropriate choice in the EML range, and keep to 60mA at higher voltage.
We can have fun debating about the driver, and DIYers can just use the one they like the look of.
But having a "recommended" design for the OT might mean that the trafo-makers like Tribute & Sowter get some business making a few sets of the same thing.
And having a few "recommended" schematics, with known-good parts makes it easy for DIYers with less design-confidence to join us, and make stuff. This approach seems to have worked wonderfully well with the "26 preamp" thread:
http://www.diyaudio.com/forums/tubes-valves/151421-26-pre-amp.html
300B is a winner here, on account of the great choice of tubes a DIYer can buy - from low-cost Shuguangs, via excellent mid-price EH and JJ, up through TJ/Fullmusic and into EML super-DHT territory.
The 300B also meets all of your design-goals:
- Linear DHT;
- can always deliver 5 clean watts;
- upgrade route through the super-300Bs for more power, or to compensate for OT losses;
- easy-obtanium (but not lo-pecuniam);
If the 307A can really survive 400V for a usable operating life, then it can probably take the same OT and power supply. Maybe it takes the place of a low-cost entry level DHT for curious experimenters, which just adds to the fun.
I know that new 300Bs cost a lot more than old-stock from Antique dealers, but I have always paid for them gladly. The JJs that I bought in 1998 worked daily for 10 years - so that's £10/year (Euro 11; USD 15/year). Pretty low-cost, as pleasure goes.
My impression of the new-model EH 300B Gold is even better - it is much better-aligned than the JJ.
As for measured curves: try comparing Audiomatica's characterisation of the Richardson (Cetron) - which was a roughly-made unloved dog -
The 300B Tube
with the new EML 300B:
EML 300B Data sheet. 22.2.08
Look very close in terms of linearity.
I believe you will find that they all measure very similarly. And provided they have correctly copied the WE internal geometry, they will. This large scale 300B geometry is far more responsible for the linear nature of the DHT, I suspect, than anything in the materials.
They all sound different, I know:
- quality of vacuum/gas. Might have given old samples an advantage, but not now, after 70 years of storage. Gas-induced grid current is noisy, the last thing you need in a headphone amplifier.
- microphony. I can vouch for JJ (older) and EH as acceptable, but it's never too good.
- Dependency on filament supply stability. EML made great fun out of the TJ/Fullmusic 300B mesh for this property. With the TJ, the anode-current goes up & down with filament voltage. This is despite its high cost, and reputation.
I developed my filament regulator on 300Bs, and one outcome was that the (old, bad quality version ) Sovtek was made to sound almost as good as the JJ, & I suspect that it had a lot to do with this effect. Filament supply must always be high quality, but with some versions of 300B, the improvements are magnified, by eliminating this problem.
Anyway, that's my submission: to meet the goals of the project, an OT with a 60mA dc current handling and 4500 to 5000 ohm primary. That way, anyone can join in, and use the same transformer. It should work with 307A, should they prove equal to the task, and will certainly work with standard 300B operating points for 320V to 400V. If the need arises for more power, just increase B+ to 450V, 500V or even 600V and use the appropriate choice in the EML range, and keep to 60mA at higher voltage.
We can have fun debating about the driver, and DIYers can just use the one they like the look of.
But having a "recommended" design for the OT might mean that the trafo-makers like Tribute & Sowter get some business making a few sets of the same thing.
And having a few "recommended" schematics, with known-good parts makes it easy for DIYers with less design-confidence to join us, and make stuff. This approach seems to have worked wonderfully well with the "26 preamp" thread:
http://www.diyaudio.com/forums/tubes-valves/151421-26-pre-amp.html
Thanks for the input/experience on this. I have no issues paying $1k for irons but $500 for a pair of tubes just hurts, so if you are having good results with the Sovtek 300B that is great news.
I agree that the 300B standard makes most sense, especially if your filament supplies "equalize" the filament design issues. That's good to hear, the lower Ra of the 300B is an advantage over the 307A's and with a 5k primary I'm sure linearity wouldn't be an issue especially with tube curves like those EML's you linked.
I agree that the 300B standard makes most sense, especially if your filament supplies "equalize" the filament design issues. That's good to hear, the lower Ra of the 300B is an advantage over the 307A's and with a 5k primary I'm sure linearity wouldn't be an issue especially with tube curves like those EML's you linked.
No, loss depends on the effective volume you are using. For a given turn ratio, primary impedance and desired primary inductance the number of turns is fixed. Lets say this is 100 for the secondary. So if we impose that half of the space is used for the secondary (because this is in theory the condition for minimum loss) having one secondary you need to use a certain wire diameter. If you make 4 secondaries then you use a smaller wire size and put them in parallel. Each secondary still has 100 turns. You might object that using a thicker wire less space is occupied by insulating material, however real diffences are negligible and small wire is a lot better at high frequency because of the skin effect. You can see a difference already at about 5 KHz!
In practice the number of secondaries depends on core size, the coil former space and actual impedances. If one starts with one secondary as a rule it could not be the best choice because space could be not used efficiently. In most cases one might need to compomise in wire size and that single secondary will result in higher DCR. If you do not use the space preserving simmetry then things will get even worse because leakage inductance and stray capacitances will be higher.
Taps are different because to get a certain secondary impendance then you will use all turns only for the higher impedance and just a fraction for the lower impendace.
Last edited:
To be exact, the 300B that I use is the 2010-year production of the EH 300B Gold-grid:
EHX.com | 300B Gold Grid - Power Vacuum Tube | Electro-Harmonix
For old-stock, I would not expect too much these days, but the old Marconi-Osram PX25A would have been ideal for this job:
http://frank.pocnet.net/sheets/060/p/PX25A.pdf
with a recommended 500V 50mA operating point, and curves that stay linear at even higher voltages than 300B!
EHX.com | 300B Gold Grid - Power Vacuum Tube | Electro-Harmonix
For old-stock, I would not expect too much these days, but the old Marconi-Osram PX25A would have been ideal for this job:
http://frank.pocnet.net/sheets/060/p/PX25A.pdf
with a recommended 500V 50mA operating point, and curves that stay linear at even higher voltages than 300B!