• 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.

My Newly Scratch-Built Tube Headphone Amplifer

Jump to Latest
#21
Tubby23 said:
But it is.
The 12:1 ratio of the sowter transformer reflects 50Ω headphones into 600Ω on the primary side - that's the whole point of using them.
For my headphones specifically, for their 70Ω, the primary sees 840Ω.
Click to expand...
Yes, but transformers transform impedance by the square of the turns ratio so your 70 ohm phones look like 70 x 12 x 12 = 10800 ohms.

I have used and measured this transformer and 12:1 is definitely the turns ratio.

Cheers

Ian
 
#23
Tubby23 said:
Of course 113v would be at 0mA.
It's a resistive loaded stage, not reactive - the transformer isn't on the plate and does not see any DC.
It's merely a coupling device/load.
Click to expand...
OK. Like I wrote, I don't really understand how you used the calculator. That could very well be because of a lack of insight on my part. But I really don't see how the way you use the loadline calculator affects the way I calculated the anode dissipation in the dc-situation (so without any ac-signal). I'm pretty sure about the validity of my calculation.
 
#26
PFL200 said:
OK. Like I wrote, I don't really understand how you used the calculator. That could very well be because of a lack of insight on my part. But I really don't see how the way you use the loadline calculator affects the way I calculated the anode dissipation in the dc-situation (so without any ac-signal). I'm pretty sure about the validity of my calculation.
Click to expand...
The calculator assumes a simple common cathode..
So I'm modelling the lower tube only with a B+ of 115.46V and 17.9mA.
Reading Blemcowe's Preamplier Book, I'm assuming the equivalent plate/anode resistor to be (2600+100)=2700R based on his Thevenin equivalent circuit.
It doesn't appear to be breaking the max dissipation rule, although i'm now undecided on what Rk values to use since the primary impedance can vary from 3600-14400 ohms depending on what heaphones i plug in and/or what impedance matching setting i use, but 10k seems reasonable as its approximately halfway between 3600 and 14400. Screenshot (142).png
 
#27
If the voltage between the anode and the cathode of a triode is 115.46 V, and the current is 17.9 mA, than the anode dissipation is 115.46 x 0.0179 = 2.07 Watt. This is the anode dissipation without any ac-signal, so the situation for dc only. For this calculation the loadline is irrelevant.

Like you, I'm not looking for a fight. As far as I know, your transformer, and the way you are using it, are fine (see post #16 by PRR on this). I just noticed that you are running your E88CC's 'very hot', that's all. I'll leave at this because I'm more or less just repeating what I already wrote before.
 
#28
PFL200 said:
If the voltage between the anode and the cathode of a triode is 115.46 V, and the current is 17.9 mA, than the anode dissipation is 115.46 x 0.0179 = 2.07 Watt. This is the anode dissipation without any ac-signal, so the situation for dc only. For this calculation the loadline is irrelevant.

Like you, I'm not looking for a fight. As far as I know, your transformer, and the way you are using it, are fine (see post #16 by PRR on this). I just noticed that you are running your E88CC's 'very hot', that's all. I'll leave at this because I'm more or less just repeating what I already wrote before.
Click to expand...
Ok, I appreciate your feedback.
 
#34
ruffrecords said:
Sounds spot on. I use 680R in my own designs. The NFB helps a lot with variable loads.

Cheers

Ian
Click to expand...
I've been crunching numbers, trying to figure out what value of Rk best serves all impedance settings:

30-100Ω @12:1 it's 4320-14400Ω.
100-250Ω @6:1 it's 3600-9000Ω
250-600Ω @4:1 it's 4000-9600Ω

With Rk=600Ω
Zl is optimized for 8600Ω
12:1 optimized for 59Ω
6:1 optimized for 238Ω
4:1 optimized for 537Ω

-------------
With Rk=620Ω
Zl is optimized for 8930Ω
12:1 optimized for 62Ω
6:1 optimized for 248Ω
4:1 optimized for 558Ω
--------------
With Rk=680Ω
Zl is optimized for 9920Ω
12:1 optimized for 68Ω
6:1 optimized for 275Ω
4:1 optimized for 620Ω

I think Rk=620Ω to be the best compromise but opinions welcome.
 
#35
Here's my Revised and expanded list:

Equation for unbypassed SSRP:

Rk=(2Rl+ra)/(u-1)

Rl=(Rk(u-1)-ra)/2

30-100Ω @12:1 it's 4320-14400Ω.
100-250Ω @6:1 it's 3600-9000Ω
250-600Ω @4:1 it's 4000-9600Ω

----------
With Rk=100Ω
Zl is optimized for 300Ω
12:1 optimized for 2Ω
6:1 optimized for 8Ω
4:1 optimized for 18Ω
----------
With Rk=360Ω
Zl is optimized for 4460Ω
12:1 optimized for 31Ω
6:1 optimized for 123Ω
4:1 optimized for 279Ω
----------
With Rk=390Ω
Zl is optimized for 4940Ω
12:1 optimized for 33Ω
6:1 optimized for 137Ω
4:1 optimized for 309Ω
----------
With Rk=430Ω
Zl is optimized for 5580Ω
12:1 optimized for 39Ω
6:1 optimized for 155Ω
4:1 optimized for 349Ω
----------
With Rk=470Ω
Zl is optimized for 6220Ω
12:1 optimized for 43Ω
6:1 optimized for 173Ω
4:1 optimized for 389Ω
----------
With Rk=510Ω
Zl is optimized for 6860Ω
12:1 optimized for 48Ω
6:1 optimized for 190Ω
4:1 optimized for 429Ω
----------
With Rk=620Ω
Zl is optimized for 8620Ω
12:1 optimized for 60Ω
6:1 optimized for 239Ω
4:1 optimized for 539Ω
-------------
With Rk=680Ω
Zl is optimized for 9580Ω
12:1 optimized for 66Ω
6:1 optimized for 266Ω
4:1 optimized for 598Ω
 
Last edited:
#36
Tubby23 said:
I've been c

I think Rk=620Ω to be the best compromise but opinions welcome.
Click to expand...
I am sure that will be fine. It is important to remember when applying these formulae that ra in particular is operating point dependent. The data sheet values are correct only for the operating point at which they were measured which is usually at higher operating current that you will normally require.

Cheers

Ian
 
#38
6mA is about right I think. 390R should give you a good margin. With SRPP is is important that the optimum load is never less than the actual load so 390R is in the right direction. I keep forgetting you have a lowish HT.

Couple of final points. Do you really need R2? What is the purpose of C14 and R46? And also the purpose of C7 and R23?

Cheers

Ian
 
#39
ruffrecords said:
6mA is about right I think. 390R should give you a good margin. With SRPP is is important that the optimum load is never less than the actual load so 390R is in the right direction. I keep forgetting you have a lowish HT.

Couple of final points. Do you really need R2? What is the purpose of C14 and R46? And also the purpose of C7 and R23?

Cheers

Ian
Click to expand...
I've been running Spice and watching the SSRP's output very carefully - the stage's push/pull is best balanced at the calculated Zload [Rk=(2Rl+ra)/(μ-1)] and becomes ever more unbalanced the further away from the calculated ideal Zload, operating less and less as a push/pull and more as a simple single ended stage.
According to Merlin Blencowe's Preamplifier book and his online "Optimized SRPP" paper.

https://drive.google.com/file/d/147K_BhXuxFv0d1c4u-RkxRZUKO2ry_gf/view?usp=drivesdk
He suggests using Ra @ 2Rk to further balance the SSRP, ie, the "pushing and pulling" at the calculated ideal Zload but it's not absolutely necessary for high μ tubes.
I can confirm this to be true from observation of Spice outputs, by comparing the signal amplitude (current) at V2's cathode just above the tap and that of Rk2, beneath it.

C14 and R46 I lifted off of John Broskie's TubeCad site - his "Akido Mojo".
It injects a portion of the Remaining B+ noise into the cathode, canceling it out.
Kind of like NFB for B+ noise.
Every time I change the RK resistor values and the quiescent current I need to recalibrate it and adjust R46.
I do this empirically in SPICE by disconnecting the PSU caps and idling the circuit while zooming in on the flatline output after the power stage's coupling cap - it's very effective at flattening the quiescent B+ even with the caps disconnected.
My amp is dead silent as it currently stands.
 
Last edited:
#40
Tubby23 said:
I've been running Spice and watching the SSRP's output very carefully - the stage's push/pull is best balanced at the calculated Zload [Rk=(2Rl+ra)/(μ-1)] and becomes ever more unbalanced the further away from the calculated ideal Zload, operating less and less as a push/pull and more as a simple single ended stage.
According to Merlin Blencowe's Preamplifier book and his online "Optimized SRPP" paper.
Click to expand...
I keep forgetting you have the advantage of a fixed load so optimisation is a reality
Tubby23 said:
C14 and R46 I lifted off of John Broskie's TubeCad site - his "Akido Mojo".
Click to expand...

OK, understood.

I use 6922 tubes in an SRPP output stage in my audio mixer designs. I have a variable load depending on what routing is selected and if the direct outputs are used or not and so on. Basically the load can vary from from moderate to heavy. My version outputs +26dBu into a 600 ohm load with a 300V supply (that is 400mW) using a 2:1 output transformer. There would typically be 20 or more such stages in a mixer quiescent current has a big impact on the power supply so it is set to be sufficient to drive the worst case load rather than to optimise the SRPP. Dissipation is also an issue with so many tubes in a smallish enclosure.

Of course Merlin would say this is exactly the application where you would NOT use an SRPP.

Cheers

Ian
 
You must log in or register to reply here.
Top Bottom