• WARNING: Tube/Valve amplifiers use potentially LETHAL HIGH VOLTAGES.
    Building, troubleshooting and testing of these amplifiers should only be
    performed by someone who is thoroughly familiar with
    the safety precautions around high voltages.

Planning first valve amp build

Having spent many years tinkering with and repairing solid state amps, the lure of building my own amp has been lingering. I originally intended to build a solid state amp from scratch but the quad 606 clone I was planning isn't really all that unique amongst what I already own. Looking for that "something different" to build has drawn me to the dark art that is vacuum tubes. That and a work colleague who has been restoring his old leak TL12's. Having spent the last week or so looking at various designs, I can see that it's quite a deep dive to get into.

Based on what I currently have and use (inter-M R300+ and Tannoy 10" super reds 91dB/W anechoic in a 9' sq room), I reckon a push pull design in the region of 20-30W is probably the most sensible. From that I came up with 3 options that are all pretty similar:

1. Mullard 5-20
2. Claus Byrith 4-30
3. Pete Millett Uniamp

Transformer wise, I've been looking at Primary Windings, VVT, Sowter, Hammond, Toroidy and Lindahl. The standout for me so far is the primary windings offerings for both the mains transformer and the output transformer as they offer a good discount when ordering two items and the specifications seem pretty good. VVT and hammond seem similar. Sowter and Lindahl are no doubt some of the best but cost a bit too much, although the lindahl LL1663 output is reasonably priced. Toroidy I'm not certain on as toroidal transformers don't seem to be common in valve outputs. Edcor can be bought in from the EU with semi reasonable shipping cost.

Valve wise, I'm thinking of sticking with JJ tesla parts that match the designs. They appear to represent good value and, based on what I've read, are supposed to perform very well.

What I'm trying to work out is whether any one of these designs is definitely the superior. From what I've read, the primary issue that I will see is one of input sensitivity. I use a marantz AV7701 preamp, currently utilising the balanced outputs. The millett amp is obviously going to be the most expensive as I have to factor in the need for both the pcbs and the sowter phase splitting transformer. I did wonder whether I could simply omit the sowter transformer as I'm already using a balanced signal but then if it were possible, I'd lose the galvanic isolation it offers. Is the uniamp design worth the ~£200 greater outlay? Is the CB 4-30 a big improvement over the standard mullard 5-20?
 
You’ve picked a good project! Some thoughts:

I get enough volume using a 7 watt Single ended triode amp and 89db speakers in a room 4x the size you mention. 20 watts will leave you with some margin. Listening habits may be different though. My next project will be a similar increase in power.

Throw Transcendar in on that list of transformer builders. I’ve bought OTs and some customs chokes from them and they’re fast and priced well.

A good but bargain price source of power transformers is Antek. They have a decent selection of toroidal on their website.

I also like JJ tubes. You can save corksniffery for after you blown some stuff up on your first build. :)

And I do think you can build the UniAmp without the transformers as long as you have differential signals, but I haven’t built it so I may be overlooking something.


Regards,
Brian
 
Last edited:
Given its "idiot resistant" nature, Mullard style circuitry is an excellent choice. However, keep in mind that Mullard intended the 5-20 to sell stuff they manufactured. The EF86 and ECC83/12AX7 are fine voltage amplifiers, but they exhibit low transconductance (gm), which can be problematic in circuits containing a GNFB loop. A superior implementation of Mullard style circuitry that employs high gm small signal types is the Harman/Kardon Citation V. You could "clone" the Cit. 5 and use inexpensive Russian 6Π3C-E (6p3s-e) [5881 equivalent] tubes to easily obtain the desired 30 WPC.

BTW, while the situation in Europe may be different, JJ's Octal production has been very inconsistent here in North America. I distrust JJ's Octals. To be fair, the situation with JJ's Noval (9 pin minature) stuff is definitely better.

P.S. You may find that line stage gain is unneeded, when a "standard" 2 VRMS CDP is the signal source.
 

Attachments

  • Citation-V-full-final.gif
    Citation-V-full-final.gif
    273.3 KB · Views: 1,068
Given its "idiot resistant" nature, Mullard style circuitry is an excellent choice. However, keep in mind that Mullard intended the 5-20 to sell stuff they manufactured. The EF86 and ECC83/12AX7 are fine voltage amplifiers, but they exhibit low transconductance (gm), which can be problematic in circuits containing a GNFB loop. A superior implementation of Mullard style circuitry that employs high gm small signal types is the Harman/Kardon Citation V. You could "clone" the Cit. 5 and use inexpensive Russian 6Π3C-E (6p3s-e) [5881 equivalent] tubes to easily obtain the desired 30 WPC.

BTW, while the situation in Europe may be different, JJ's Octal production has been very inconsistent here in North America. I distrust JJ's Octals. To be fair, the situation with JJ's Noval (9 pin minature) stuff is definitely better.

P.S. You may find that line stage gain is unneeded, when a "standard" 2 VRMS CDP is the signal source.
Good shout on the Citation V. Having had a look on Hfe, there is a full assembly manual on there which is very helpful. Only thing that's not so easy to figure out are the transformers. I can see it appears to have what looks like a 400V secondary, a 6.3V heater secondary and a 60V secondary. There is no info I can find on the current ratings of the secondaries. For the output transformer, I have found references to using a primary resistance of 6600ohms but not much else. These niggles might require custom transformers to be made unless anyone knows of existing ones that fit the bill.
 
Most of brand schem dont work.
You can order power and output transformers from EDCORUSA.
For PT You need not less 250 - 270 ma and about 6 A - 6.3 v.
For OTs - PP, not less 60-70 ma. For 60 V- no more 20-30 ma /see links for ordering/. You will wait 40 days.
Good luck!
 
Kei, have you decided yet?

With your sensitive speakers and small room 15wpc should be lots. Forget Edcor, it will cost more for shipping than the trannies. The Tubelab SPP (simple push-pull EL84 amp) is a sweet little amp. I just built one using old Heathkit iron and it sounds good. Likely even better with new trannies. See my build in the link.

Primary windings in the UK look decent and shipping won't be a killer. Maybe go for the 25 watt 6,600 ohm push-pulls 4 quid less each rather than their 10 watt Mullard 5-10 clones.

Their 5-10 stereo power trans would suit this amp just fine.

I used a 1.5H 200Ma Hammond choke in mine and it killed the PS buzz.

Cheers, Steve


SPP Build
 
What about one like these? 22 Compactron DC coupled plus choke input filter and damper diode bridge? No one SS device at all? The final 6JN6's are in series and autotransformer coupled to the speaker(s). Aslo positive current feedback to reduce output impedance near zero ohms. Still unfinished.
 

Attachments

  • Amp 204.JPG
    Amp 204.JPG
    146.6 KB · Views: 1,024
  • Amp 205.JPG
    Amp 205.JPG
    144.2 KB · Views: 991
  • Amp 206.JPG
    Amp 206.JPG
    138.8 KB · Views: 966
Good shout on the Citation V. Having had a look on Hfe, there is a full assembly manual on there which is very helpful. Only thing that's not so easy to figure out are the transformers. I can see it appears to have what looks like a 400V secondary, a 6.3V heater secondary and a 60V secondary. There is no info I can find on the current ratings of the secondaries. For the output transformer, I have found references to using a primary resistance of 6600ohms but not much else. These niggles might require custom transformers to be made unless anyone knows of existing ones that fit the bill.


The Cit. 5 B+ supply employs "full wave" voltage doubler topology. So, the RMS voltage of the B+ winding is much lower than 400 V.

6.6 Kohm primary O/P "iron" is easy to obtain. FWIW, the Cit. 5's OEM stuff uses a lower primary impedance to squeeze more power out of 7581s. However, your speakers will do just fine with the less powerful / less costly Russian 5881 equivalents. The 6L6 family of tubes has been in continuous production since RCA introduced the type in the 1930s. 80+ years and going strong.

Given their weight, magnetics are damned expensive to ship. Manufactured reasonably close to home makes sense. UK based Sowter makes a quality product. Their model U069 O/P trafo will be fine. Magnetic headroom is essential in a design that incorporates a GNFB loop. That headroom avoids core saturation caused by the deep bass error correction signal. A transformer with a saturating core sounds truly horrid.

The Cit. 5 employs full pentode mode "finals". Open loop linearity in full pentode mode is maximized, when screen grid (g2) B+ is regulated at a % of anode B+. Do something to stabilize g2 B+ beyond the simple RC decoupling employed by H/K. Jim McShane, a Citation "guru" and all around very nice guy, uses a choke.
 
Most of brand schem dont work.
You can order power and output transformers from EDCORUSA.
For PT You need not less 250 - 270 ma and about 6 A - 6.3 v.
For OTs - PP, not less 60-70 ma. For 60 V- no more 20-30 ma /see links for ordering/. You will wait 40 days.
Good luck!
The photofacts data says Pri = 117V 2.1A, Sec 1 = 200V 1.2A, Sec 2 = 6.3V 6.3A, Sec 3 = 60V 0.02A. Those figures would suit a transformer that would be used for the original stereo design. My plan is to build mono blocks so one transformer per unit so approximately half the current rating would make sense and obviously I'd need a primary that suits our local 240V 50Hz UK supply.
The Cit. 5 B+ supply employs "full wave" voltage doubler topology. So, the RMS voltage of the B+ winding is much lower than 400 V.

6.6 Kohm primary O/P "iron" is easy to obtain. FWIW, the Cit. 5's OEM stuff uses a lower primary impedance to squeeze more power out of 7581s. However, your speakers will do just fine with the less powerful / less costly Russian 5881 equivalents. The 6L6 family of tubes has been in continuous production since RCA introduced the type in the 1930s. 80+ years and going strong.

Given their weight, magnetics are damned expensive to ship. Manufactured reasonably close to home makes sense. UK based Sowter makes a quality product. Their model U069 O/P trafo will be fine. Magnetic headroom is essential in a design that incorporates a GNFB loop. That headroom avoids core saturation caused by the deep bass error correction signal. A transformer with a saturating core sounds truly horrid.

The Cit. 5 employs full pentode mode "finals". Open loop linearity in full pentode mode is maximized, when screen grid (g2) B+ is regulated at a % of anode B+. Do something to stabilize g2 B+ beyond the simple RC decoupling employed by H/K. Jim McShane, a Citation "guru" and all around very nice guy, uses a choke.
The sowter would be lovely but the cost is somewhat hard to justify at this stage. My preferred budget for the whole build is around £5-600. The cheaper the better provided I'm not seriously sacrificing the quality.

Kei, have you decided yet?

With your sensitive speakers and small room 15wpc should be lots. Forget Edcor, it will cost more for shipping than the trannies. The Tubelab SPP (simple push-pull EL84 amp) is a sweet little amp. I just built one using old Heathkit iron and it sounds good. Likely even better with new trannies. See my build in the link.

Primary windings in the UK look decent and shipping won't be a killer. Maybe go for the 25 watt 6,600 ohm push-pulls 4 quid less each rather than their 10 watt Mullard 5-10 clones.

Their 5-10 stereo power trans would suit this amp just fine.

I used a 1.5H 200Ma Hammond choke in mine and it killed the PS buzz.

Cheers, Steve


SPP Build
I've not made any decisions yet. I agree that on the power front, I don't think I need more than 15-25W which is where I was aiming. A quick measure on my scope last night at moderately loud listening levels, I was seeing 1.44V rms, 5.8V P-P at most. Obviously having some headroom above this makes sense as it wan't as loud as I have had in the past, but it was a bit above the norm. I do quite like the look of the Citation V that Eli suggested and I'd prefer to work P2P as it reduces the cost and makes future modification so much easier. I just have to take care when assembling it all. The great attractiveness to the original mullard design is the ease of obtaining the components with no need to delve into custom would transformers and the greatly detailed assembly instructions. Also, don't forget that not only is this my first foray into the world of vacuum tubes, it will also be my first scratch build amplifier. (Ignoring the LM386 I built 18 years ago)
 
And now for something completely different......

Kei,
The Mullard 5-10 probably has too much gain for modern sources, it's that EF86 pentode. Have a look 'round the Tubelab.com forum. The schematic for the TL SPP is there somewhere. It could easily be built P to P. The alternative is, though it would be less powerful, is one of the Tubelab SE amps.

The thing is, though the Citation clone could be nice, you'd need some decent (read: fairly pricey) OPTs to get the most out of it.

Have you been following forum member Soundhappy's posts of the Japanese DIY articles? I've accumulated some spare trannies and I'm in the process of building a Quad-like EL84 PP amp, just for the hell of it. See attachment.

Steve
 

Attachments

  • MJ1405-075_EL84-PP-power-amp_IwamuraY.pdf
    1.6 MB · Views: 127
My take on this is go simple for your first build especially as you want it to be a scratch build. Choose something that you can get parts for easily (obviously) that is well documented and widely understood. Use it as practice, not the final product...

As you are UK based, the Mullard designs are ideal. You can get transformers without shipping and import tax. They are not perfect, as noted, but treat your first attempt as learning 'how to' and 'what I want' ready for the next build.
 
A superior implementation of Mullard style circuitry that employs high gm small signal types is the Harman/Kardon Citation V. You could "clone" the Cit. 5 and use inexpensive Russian 6Π3C-E (6p3s-e) [5881 equivalent] tubes to easily obtain the desired 30 WPC.


I guess the Citation V's PI stage's performance could even be improved by increasing the tail resistor to 39 kOhms and returning it to the -54Vdc rail instead of GND.
Best regards!
 
I guess the Citation V's PI stage's performance could even be improved by increasing the tail resistor to 39 kOhms and returning it to the -54Vdc rail instead of GND.
Best regards!


Or replace the tail resistor with a CCS, IXYS 10M45S.
But I agree the Cit 5 is a nice circuit.
Not 100% sure there couldn't be a better choice than the 12BY7 in front though
 
Or replace the tail resistor with a CCS, IXYS 10M45S.
But I agree the Cit 5 is a nice circuit.
Not 100% sure there couldn't be a better choice than the 12BY7 in front though

Changing the LTP tail load to a 10M45S is a good idea. It may be necessary to add a small negative rail, to ensure "compliance".

The high gm theme can be carried further by using an ECC99, instead of a 6CG7, as the LTP.

It's very hard to argue with anything Stu Hegeman did. The big issue (IMO) with the 12BY7 is ever increasing scarcity. I'm providing a cascode replacement for the 12BY7 pentode voltage amplifier, that employs the in production 6922.
 

Attachments

  • Cascode 6922 Mullard Front End.jpeg
    Cascode 6922 Mullard Front End.jpeg
    636.5 KB · Views: 384
SPP point to point

It is not very easy buidling a P-P amplifier as a first project, IMO. You have to consider the orientation of the tubes, and the routing of wiring that could be sensitive to noise.

Once the heaters are wired, it is not very easy to change anything, and that means you have to have a good idea how the terminal strips will be located, and how the larger components will be positioned.

My approach was to start with a Tubelab SPP. Then I cloned that design so that I could make use of cheap PL84's (15v heater EL86's). I built the identical circuit P-P into a rather natty baking tray I found in recycling, so that I could use the working circuit as a reference, and so that I could think about the spatial parts of the build. (Note the clutter free and organised working environment).

Alternaively, there are builds based on turret board that can be cloned, or do it on a breadboard, so that the tube holes are not cut in your chassis, limiting your options, allowing re-work.
 

Attachments

  • IMG_1359.JPG
    IMG_1359.JPG
    859.7 KB · Views: 342
  • IMG_1361.jpg
    IMG_1361.jpg
    349.7 KB · Views: 333
Member
Joined 2004
Paid Member
Look into "Toroidy" of Poland. Magnetics are the the "big ticket" items, when tubed circuitry is being constructed.

+1 for Toroidy.

They have a very reasonably-priced PP (optional UL) output transformer with 6600 ohm primary that was discussed and tested in the “Baby Huey EL34” to very good reviews, should you choose to build a Citation V as Eli recommended using 5881 type output tubes (About US$134 per pair from TME at the moment). Also 8000 and 10k Ohm PP UL outputs if you decide EL84/6V6 would provide enough juice.

See: TTG-EL34PP TOROIDY - Transformer: speaker | 50VA; O115x65mm; 0.008/54kHz; 200mA | TME - Electronic components

I bought several pairs of the PP transformers from TME; service was great to my Mid-America location. A pair of EL34 transformers shipped from Poland to me for US$9.90, reached me within 3-4 days of ordering. Be aware shipments exceeding 5 kg are significantly more expensive, so buy one-at-a-time if your choice exceeds 2.5 kg per transformer, like the KT88PSE ones George ordered. (I have no connection with Toroidy other than being a satisfied customer.)

BTW, George, TME now has many 3k primary KT88SE transformers in stock. Impatience wil get you sometimes.