• 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.

Seeking Critique of first Tube Amp Design

kodabmx

Member
Paid Member
2011-10-31 1:00 am
Toronto
They glue cars together these days. Adhesives work well if they are designed for the job at hand.

Believe it or not, I've used hot glue to mount things like relays in a hot tube amp. I don't use it anymore but it does work.

Cyanoacrylite (Krazy Glue) works well, too.
 

GregH2

Member
2009-10-30 1:43 pm
Hi all,

I now have the chassis more or less complete with just some holes for the FETs remaining. Before I drill, I'm wondering if my driver is overkill. It seems a simple common cathode stage with shunt feedback from anode to anode might be the way to go - obviously with a different octal tube, maybe a 6SL7 running at 2mA or so.

Something like this:

6SQ7-EL34-Single-Ended-Tube-Amp-Stephe-Final-Version-2.png | EL34-6SQ7 Amp

I've been sucked in by the subjective reviews surrounding this design and a similar one at blueglow.

Any thoughts? Should I stick with the follower? I used it initially because I thought it would give pretty good performance in a zero GFB design, but now I am not so concerned about using feedback. I also thought it might work better if I run a larger tube like a KT88.

I guess if I am seeking better objective performance I could add shunt feedback to the follower design later...

Appreciate any thoughts.

Greg
 

GregH2

Member
2009-10-30 1:43 pm
Hi All,

Well, my first tube amp is complete and it sounds rather wonderful! The amp has a lovely dynamic and engaging sound, particularly with CFB and U/L enabled, I think due to the high bandwidth transformer and source follower driver. Sounds wonderful to my ears despite poor objective performance!

Thank you all for your help over the last few months.

Measured THD in triode mode with zero feedback @ 1 Watt into 8R is about 2%. Applying local cathode feedback to the output stage drops this by about 25%. Running in U/L is much worse. I have tried alternative input and output tubes with only minor change. Both channels measure more or less the same. Sonically, I really like the gold lion KT77 even though it's not the best measuring. Frequency response with no feedback is down 1 dB at 30Hz and 40kHz into 8R.

Now, I always knew a single ended tube amp with no global feedback was going to be high distortion, but this seems materially worse than other published designs. My suspicion is that this is because the input stage and driver have such low distortion that there is no cancellation with the output stage. While I am not caught up on measured figures, I would like the best performance achievable for the topology. I have posted the measured operating points below and would appreciate any feedback regarding possible improvements or glaring errors.

I am running the output stage right on its limit, is this potentially overdoing it? Was thinking a KT88 might be more comfortable here, with perhaps a tad lower cathode resistor. Thoughts?

If the design is more or less maxed out so be it, I am enjoying the sound as is.

Cheers,

Greg
 

Attachments

  • SE_EL34_Amp.jpg
    SE_EL34_Amp.jpg
    306.9 KB · Views: 94
  • Amp Operating Points.PNG
    Amp Operating Points.PNG
    443.8 KB · Views: 98
Last edited:
Is that sand cathode follower driver setup needed? I find with SE amps, simple is better. An EL34 isn't a hard tube to drive.

The other question, in UL mode, how much resistance do you have between the UL tap and the screen? I like 1.2-1.4k which helps drop the voltage on the screen some. if you are only using 100R, you might be running more screen voltage than plate voltage. You are running that tube pretty hard but I like the sound of SE amps with the tubes run hard, I don't think that is a problem.
 

GregH2

Member
2009-10-30 1:43 pm
Hi Stephe,

Thanks for taking the time to respond with some thoughtful comments and love your channel!

First, the screen resistor...

Yes the screen resistor is 100R, but I wouldn't have thought this would be an issue in triode mode? However, I can see how perhaps in U/L it is an issue, so I will try something larger this weekend and get back to you. Thanks for the suggestion!

Now, the source follower:

I used a source follower because:

1) A source follower used in this way typically has fractions of a percent distortion, so I do not expect it has any detrimental effect on the sonic qualities of a tube amp with several percent distortion. The distortion it provides is likely swamped. Certainly on the FFT from my audio analyser I can't see any measurable high order artifacts.

2) I wanted the option of running larger KT88 or KT120 tubes (with a re-bias). Anecdotal reports I have read suggest these larger tubes sound anaemic if not properly driven. As it is, the amp has lots of drive and is flat out to well beyond 30kHz with no feedback, so I am quite happy with the driver. No issues driving the input capacitance here!

2) The source follower is bootstrapped which provides a constant current source for the 6SN7. This extracts maximum gain at minimal distortion from the tube and allowed me to use a single 6SN7 as the input and still have enough gain.


Despite the high THD it is amazing how dynamic and tight the amp sounds, it's quite amazing compared to other singled ended amps I have heard.

Cheers,

Greg
 

wg_ski

Member
2007-10-10 5:21 pm
Since the 4.7 uF capacitor is what is responsible for the k-follower stage to add gain, I’d say it converts it to a hybrid amp. Without the bootstrap the gain would drop a smidge because the 36 k would appear in parallel with Ra, so in effect the follower is adding gain. If it works and you like it, you can just call it a “success”.
 
Since the 4.7 uF capacitor is what is responsible for the k-follower stage to add gain, I’d say it converts it to a hybrid amp. Without the bootstrap the gain would drop a smidge because the 36 k would appear in parallel with Ra, so in effect the follower is adding gain. If it works and you like it, you can just call it a “success”.
Yes, this is what I meant, but I didn't express myself very well.
 

GregH2

Member
2009-10-30 1:43 pm
Yes, as Tom says I thought the bootstrapped fet was a neat way of adding gain and driving the output with a low impedance...without the need for a elevated heater supply or active CCS in the input stage.

You can read more about the concept here, at the bottom of the page, albeit for a cathode follower:

http://www.valvewizard.co.uk/dccf.html
I am not worried about using a Fet this way as the minute sonic influence it may have is likely swamped by that of the tubes. This is much like tubelabs powerdrive concept which many seem to rate highly.

One negative side effect, in simulation at least, is that the overall distortion of the amp is much higher because the input stage offers very little of its own distortion to counteract that of the output stage. So perhaps I should at least give it a go without it, although I may need a different input tube given the decrease in gain that will result.
 
Hi Stephe,

Thanks for taking the time to respond with some thoughtful comments and love your channel!

First, the screen resistor...

Yes the screen resistor is 100R, but I wouldn't have thought this would be an issue in triode mode? However, I can see how perhaps in U/L it is an issue, so I will try something larger this weekend and get back to you. Thanks for the suggestion!


Greg
100R is fine for triode strap but IMHO is too low for UL tap and might be why you say UL sounds much worse. I aim to get the screen voltage at least a few volts below the plate. With some tube types lowering the voltage even more can help reduce distortion.
 

GregH2

Member
2009-10-30 1:43 pm
100R is fine for triode strap but IMHO is too low for UL tap and might be why you say UL sounds much worse. I aim to get the screen voltage at least a few volts below the plate. With some tube types lowering the voltage even more can help reduce distortion.
Hi Stephe,

Thanks for the comments. Tonight I tried your suggestion and upped the screen resistor to 1.1k. Unfortunately it didn't have any material impact on the distortion figures in ultra linear mode.

Cheers,

Greg
 
Last edited:
Hi Stephe,

Thanks for the comments. Tonight I tried your suggestion and upped the screen resistor to 1.1k. Unfortunately it didn't have any material impact on the distortion figures in ultra linear mode.

Cheers,

Greg
What is the screen vs plate voltage? And yes, you probably need some NFB in UL mode, I like the plate to plate "Schade" feedback myself on a SE amp. Something in the 220k-470k is a good starting point, and the value will affect the tone, higher values preserve the highs, lower values tamp tamp them down.
 

GregH2

Member
2009-10-30 1:43 pm
What is the screen vs plate voltage? And yes, you probably need some NFB in UL mode, I like the plate to plate "Schade" feedback myself on a SE amp. Something in the 220k-470k is a good starting point, and the value will affect the tone, higher values preserve the highs, lower values tamp tamp them down.
Thanks Stephe,

I didn't actually think to measure it! Will have another go at it this weekend and get back to you.

I'm currently using some cathode feedback which drops the distortion in triode mode by about 25% and about 50% in U/L - clearly more gain in U/L with which to provide feedback. Will give the Schade feedback a go.

With all these modifications I am really wishing I had made my layout more modification friendly!

Cheers,

Greg
 

GregH2

Member
2009-10-30 1:43 pm
Hi Stephe.,

So... bit of a mish mash of progress and a quick update. I measured the screen voltage with just the 100R screen stopper in place. At idle it was pretty much equal to the anode voltage, maybe 0.5V above. So I installed an additional 1k2 screen resistor which means I now have 1k3 on the screen in U/L mode. Silly me forgot to actually measure what the screen voltage was again, but presumably it is well below the anode now. I did measure distortion which was more or less unchanged, maybe marginally better in U/L mode than previously.

Next, I installed some KT88's with the cathode resistor unchanged. They are running at about 70mA and nearly 30W, not ideal for a tube this big but about the limit of my power supply xformer which is now getting very toasty. This has completely transformed the amp! Distortion is down to about 1% at 1W and is actually better in U/L than triode, so U/L is working as it should. The amp has so much kick with the KT88's it's not funny, bass performance is great and highs are very clean. Do I like the sound more though? Probably not, I think the EL34 sound is nicer despite the much higher THD and flabbier bass. So there you go.... more to sonic enjoyment than THD alone.

With all tubes tried to date I prefer U/L mode with local cathode feedback on the output tube. Triode is probably nice for simple music but U/L is more versatile.

Cheers,

Greg
 
Maybe if they are active devices... Passives like diodes get a pass IMHO.
For prtection, maybe, but for bias like LEDs up cathodes, no.
I would extend that to mean any SS device that isn't varying with signal, too. like a CCS?
You know you can make a great CCS with a pentode.

Cheers

Ian