I'm currently getting all the parts together to make a valve amp and I am a little stuck on the output transformers.
The specs say to use 6600ohm push-pull transformers rated at 30-40w.
Here in Japan, it's very difficult to find that value. 5000 and 7000 ohm are everywhere though (I can literally buy about three or four different brands of them off the shelf from multiple shops in Akihabara...)
I have 7591 output tubes, but I'm not sure if I can use the 5000 or 7000ohm transformers and what modifications I'll need to make to use them.
Can someone educate me? I'm new to all this tube amp stuff. 🙂
The specs say to use 6600ohm push-pull transformers rated at 30-40w.
Here in Japan, it's very difficult to find that value. 5000 and 7000 ohm are everywhere though (I can literally buy about three or four different brands of them off the shelf from multiple shops in Akihabara...)
I have 7591 output tubes, but I'm not sure if I can use the 5000 or 7000ohm transformers and what modifications I'll need to make to use them.
Can someone educate me? I'm new to all this tube amp stuff. 🙂
7 Kohms will be fine.
If the circuit you plan on using places the O/P "iron" inside a NFB loop, get stuff rated for 60 to 70 W. You need magnetic headroom to deal with the deep bass error correction signal.
If the circuit you plan on using places the O/P "iron" inside a NFB loop, get stuff rated for 60 to 70 W. You need magnetic headroom to deal with the deep bass error correction signal.
Yes, the tube manuals say to use 6600 ohms for a general purpose push-pull output stage with 7591A's. But this is a design center value and some variation is allowed. Since 7000 ohms is quite close, I would use that one. In general a lower impedance match will provide slightly more power along with a little more distortion. Conversely a higher match will give slightly less power with a little less distortion. These differences are small but measurable. The difference between the 6600 and 7000 will be difficult to hear if at all. No damage or strain will will occur from using a little higher or lower primary impedance match. For the 7000 ohm transformer no modification is required.
Awesome.
It's only about a 6% change, so well within the usual 10% tolerance of most parts anyway.
I thought a small change would be acceptable, but I wanted to make sure before I cooked anything...
It's only about a 6% change, so well within the usual 10% tolerance of most parts anyway.
I thought a small change would be acceptable, but I wanted to make sure before I cooked anything...
Also, remember that the primary impedance will fluctuate based on influence from the output load/tap interface being reflected back. No speaker I know of is flat across the board with regard to impedance. IMHO best to go with 7k for best distortion characteristic.
FWIW, I'm not impressed with that design, as it appears to be needlessly complex. The only thing I like is regulated 7591 screen grid B+. Net gain rates to be excessive, for use with modern, digital, sources. OTOH, the basic topology of voltage amplifier DC coupled to a "concertina" phase splitter and on to the "finals" is valid. That style of circuitry was used by many manufacturers of note. You can snip circuitry fragments from several old and not so old designs to get a very nice end result.
The O/P transformer is inside a global NFB loop and, as stated previously, 60 + W. rated "iron" is in order. Do the transformers you plan on using have ultra-linear (UL) taps?
High PIV Schottky diodes are every bit as quiet as vacuum rectifiers. IMO, there is very little reason (if any) to use a pair of GZ34/5AR4s and the "baggage" they come with in a new power amp build.
The O/P transformer is inside a global NFB loop and, as stated previously, 60 + W. rated "iron" is in order. Do the transformers you plan on using have ultra-linear (UL) taps?
High PIV Schottky diodes are every bit as quiet as vacuum rectifiers. IMO, there is very little reason (if any) to use a pair of GZ34/5AR4s and the "baggage" they come with in a new power amp build.
Yeah, Schottky diodes are on the cards for the new BOM (probably from Cree), as are UF5408 diodes in place of the selenium of course, and a PTC on the input of the power transformer, the usual modern upgrades.
Should I make any changes to the rest of the circuit to add the Schottky diodes, or can I stick them in and adjust the bias values etc to suit?
I'm still new to this stuff, so what sort of things would you suggest changing in particular? You might have to spell it out slowly, I really want to learn this stuff!
Should I make any changes to the rest of the circuit to add the Schottky diodes, or can I stick them in and adjust the bias values etc to suit?
I'm still new to this stuff, so what sort of things would you suggest changing in particular? You might have to spell it out slowly, I really want to learn this stuff!
This design has clearly been optimized at high frequencies for the original OPT. It's safe to ignore a small gain difference due to the 7K vs 6.6K primary impedance difference, but OPT substitution frequently introduces BIG changes in leakage inductance, stray capacitance, self-resonance frequency and Q, etc. You will probably be compelled to re-optimize the values of C1, C2, C3, C4, R7, and R8. This calls for a squarewave generator, scope, and dummy loads, as well as some skill at the test bench.
Hmmm, sounds like a good way to mess things up and then learn how to fix it. 🙂
Maybe I'll start first by building the amp basically as shown re. topology, then once it's working I can sit down with my scopes and stuff (I'm well setup for digital electronics) and start poking where I shouldn't and see what happens. 🙂
I guess I could poke one channel and compare it to the unmodified channel as a reference to see what makes an improvement and what doesn't.
Maybe I'll start first by building the amp basically as shown re. topology, then once it's working I can sit down with my scopes and stuff (I'm well setup for digital electronics) and start poking where I shouldn't and see what happens. 🙂
I guess I could poke one channel and compare it to the unmodified channel as a reference to see what makes an improvement and what doesn't.
Attaboy!
It might oscillate vigorously on your first attempt. If so, you will need to open the feedback loop and then add feedback incrementally as you adjust the HF tweaks. Check it without load and with some capacitive load as well. It should be free from oscillation up to 0.1uF at least, but it's more-or-less normal to see some overshoot and ringing with a hard C load.
Eico ST 70 uses 7591 and the OPT is 6K6 . I've a pair on my shelf for years.
If, interested, message me!
Tamura F-684
I would not always trust the original design values....
I believe the 6.6K was chosen mainly because it was a very common plate load back in the day ...
Also because the Sylvania 7591 data sheets recommends it at those voltages to make 45W..
Examining the plate load for 400V screen...
The plate load choice depends on how solid the plate voltage will be during full power output....ie how much of a dip the plate voltage go down from 475...
Assuming your plate voltage will dip to 425V, a 5K plate load looks like a good choice...
Assuming a droop to 450V...still your under 6K for nice fit plate load...
Calculation comes up exactly 50W...
The 6.6K is a bit safer for average plate dissipation on the tubes during AC operation..
I believe the 6.6K was chosen mainly because it was a very common plate load back in the day ...
Also because the Sylvania 7591 data sheets recommends it at those voltages to make 45W..
Examining the plate load for 400V screen...
The plate load choice depends on how solid the plate voltage will be during full power output....ie how much of a dip the plate voltage go down from 475...
Assuming your plate voltage will dip to 425V, a 5K plate load looks like a good choice...
Assuming a droop to 450V...still your under 6K for nice fit plate load...
Calculation comes up exactly 50W...
The 6.6K is a bit safer for average plate dissipation on the tubes during AC operation..
Last edited:
- Status
- Not open for further replies.