[L0rdGwyn]

HiLG,
Very interesting to read about all the particulars in building your amp, but not having any electronic background I can't say that I understand much about what you are describing.
Apart from the tubes you are using, is this a conventional type tube amp or an innovative design?

Hey mordy - I'm sorry, it is pretty technical, maybe at some point I can try to make it a little more easy to understand.

This is a conventional design, a single-ended triode (SET) topology, also known as class A. The aspects of it that are somewhat unconventional are the choice of driver tubes and the cathode bias resistor switch for multiple types of output tubes being run as strapped triodes rather than pentodes/beam tetrodes. But overall, SET is a very well-known topology, good for a beginner

 

[2359glenn]

I think I might have to look into having some custom chokes wound from Heyboer @2359glenn . To get the MH4 up to >3mA on the plate like @SonicTrance suggested, need more B+ and I'm losing too much voltage on the chokes with the current draw of the power tubes. Could go to 4H 65ohm 225mA but not great losing the filtering, think I end up with 14mV ripple on the power tubes.

Running everything at a higher voltage will increase the linearity of the MH4, but the linearity of the EL34 bias point will suffer. Going to have to find a balance, but will favor the MH4.

Man, you change one thing in this business and everything else changes too! Tricky tricky.

Edit: I might have to bite the bullett and lower my plate resistor. Was hoping to keep 56kohm since I have these Riken resistors, but the tubes will be biased better at 30-40kohm.

I can send you a couple of nice 33K ohm Caddock resistors if you want them.
You are finding out everything is a compromise you have to do what affects the sound the least. I wouldn't reduce filtering in the power supply.
You could try using a constant current source for the plate load on the MH4. In my EL3N amp I use a CCS on the plate of the EL3N driver.
The same problem not enough B+ voltage plus the CCS rejects power supply noise. You can eliminate the last R-C in the power supply.
You can get a cascode CCS kit for cheap from K&K audio for $17.
That is what I use in the EL3N amp.
It is two more circuit boards with heat sink that you will have to squeeze under the chassis.

 

[L0rdGwyn]

I can send you a couple of nice 33K ohm Caddock resistors if you want them.
You are finding out everything is a compromise you have to do what affects the sound the least. I wouldn't reduce filtering in the power supply.
You could try using a constant current source for the plate load on the MH4. In my EL3N amp I use a CCS on the plate of the EL3N driver.
The same problem not enough B+ voltage plus the CCS rejects power supply noise. You can eliminate the last R-C in the power supply.
You can get a cascode CCS kit for cheap from K&K audio for $17.
That is what I use in the EL3N amp.
It is two more circuit boards with heat sink that you will have to squeeze under the chassis.

And here I was thinking I was done with the design phase. I think I could get away with a plate resistor load if I run the input and output tubes around 300V, would get me around 3.6mA on the MH4 plate with a 36kohm load. Let me look into using a CCS load, those boards are small I could probably fit them in too, it's a decent sized chassis. Might as well take my time and get this right.

Kind of ironic, these tubes are from the 1930s and yet they might have PCBs attached to nearly every pin lol.

Edit: I think I am going to do it, after I make sure I can fit two sets of boards.

Great linearity with a 6mA CCS. Max plate dissipation of the MH4 is 2.5W, should be no problem. Here is the E424N load line too at 6mA.

 

[2359glenn]

Hi Keenan

Now you learned something!! Not so easy to get it right. CCS are awesome and sound great.
Now you have a very old tube with transistors attached to the plate the horrors .
I know of no commercial amps that use CCS. Other then DIY only @SonicTrance and me use them as far as I know.

Just letting you know the CCS has a pot to set the current plus the board has a place for a fixed resistor.
After getting the right current measure the pot and get the equivalent resistor. There is a jumper to cut to eliminate the POT.
I have had the pot go noisy on people's amps in the past. Now I just put the resistor that has been measured for a EL3N.
You will also need a 10 ohm resistor in series to measure the current through the CCS

 

[tomb]

There are other instances of commercial and DIY amps that use CCS. ECP Audio has used them for years.

 

[L0rdGwyn]

Hi Keenan

Now you learned something!! Not so easy to get it right. CCS are awesome and sound great.
Now you have a very old tube with transistors attached to the plate the horrors .
I know of no commercial amps that use CCS. Other then DIY only @SonicTrance and me use them as far as I know.

Just letting you know the CCS has a pot to set the current plus the board has a place for a fixed resistor.
After getting the right current measure the pot and get the equivalent resistor. There is a jumper to cut to eliminate the POT.
I have had the pot go noisy on people's amps in the past. Now I just put the resistor that has been measured for a EL3N.
You will also need a 10 ohm resistor in series to measure the current through the CCS

Not just transistors on the plates Glenn, the filaments too! Will I ever be forgiven?

Great, I'll order the boards when I get the transformers from K&K. Using these and the filament supplies is going to make actually building the amp a lot easier, fewer components to mount to the walls of the chassis, don't have to worry about rectifiers and filter caps for the MH4 filaments and no RC filter.

One thing I am wondering and haven't been able to find information on yet: how does a CCS load affect the output impedance of the driver stage? I am thinking mostly about the low pass RC filter created by the output impedance of the driver stage and the input capacitance of the output stage. With the 56kohm plate resistors I was planning on using and the input capacitance of the EL34 (with Miller capacitance), the high frequency roll off I estimated was well over 20kHz.

How does a CCS load affect this roll-off point?

Edit: For the sake of sparing those who are subscribed from a million notifications, I am consolidating posts. Some new tubes arrived today, these are Opta Loewe REN904. Interesting boxes, very different from others of the time period.

 

[leftside]

My 1101 Audio amp uses CCS.

 

[SonicTrance]

I started messing around with LTSpice the other day, but once I got started I realized the learning curve was going to require a lot of time, so I shelved it. Definitely going to try it again when I have time!

Yeah, it can seem like a bit much at first but is very useful once you learn the basics. You can also simulate power supplies with LTspice and make triode curves for pentode tubes if they're not displayed in the original datasheet.

Man, you change one thing in this business and everything else changes too! Tricky tricky.

Yes! There're a lot of variables to think about!

Kind of ironic, these tubes are from the 1930s and yet they might have PCBs attached to nearly every pin lol.

Oh no! SS parts in a tube amp? Are you crazy or what?? Haha!

Kidding aside it looks great! Nice work on the load lines. To be fair though you've drawn the CCS load lines too far. They should end @ B+ minus maybe 10V or so as the CCS would need some headroom.

One thing I am wondering and haven't been able to find information on yet: how does a CCS load affect the output impedance of the driver stage?

The CCS wont affect that impedance much. The plate load is there to set a load line for the tube, that's it. You need to look at the tubes internal resistance called Rp or Ra. The Rp stated in datasheets are for a specific operating point. It fluctuates some as well. There's a way to calculate the Rp of a specific operating point by looking at the grid curves. Remember, it's all about the curves!
There's also this formula that's useful.

µ = gm * Rp

 

[A2029]

How does a CCS load affect this roll-off point?

Edit: wait I think I got it! The effective output impedance of the tube will be the plate voltage divided by the constant current. So if biased for ~175V on the plate with 6mA of constant current, it would be equivalent to a 30kohm plate resistor load. Does that sound right? That would be no problem for high-frequency rolloff.

To get that same bias point using a resistive load would require a B+ of about 355V, but with the CCS I can get it with the 250V I want on the power tubes so I will be get betting a higher plate current, better frequency resposne, and noise rejection. What a huge advantage, I am sold.

I'll point you to some reading on CCS's:
https://www.bartola.co.uk/valves/2015/08/31/ccs-not-everything-that-glitters-is-gold-part-i/
https://audioxpress.com/assets/upload/files/Sources_101_P1.pdf
https://audioxpress.com/assets/upload/files/Sources_101_P2.pdf

A good way to lower the output impedance of a CCS is to use the Mu output, instead of the plate output:
https://www.bartola.co.uk/valves/2018/04/28/hybrid-mu-follower-output-impedance/
https://www.audioasylum.com/cgi/vt.mpl?f=tubediy&m=182689&VT=T


The cascode mosfet CCS that K&K sells uses the IXTP01N100D as the top device and a DN2540 as the bottom device. Using those devices in a cascode you get massive PSRR and AC impedance (up to 500Mohm or more at low frequencies!). Take the Mu output off of that board and you get very low Zout as well

 

[L0rdGwyn]

Yeah, it can seem like a bit much at first but is very useful once you learn the basics. You can also simulate power supplies with LTspice and make triode curves for pentode tubes if they're not displayed in the original datasheet.


Yes! There're a lot of variables to think about!


Oh no! SS parts in a tube amp? Are you crazy or what?? Haha!

Kidding aside it looks great! Nice work on the load lines. To be fair though you've drawn the CCS load lines too far. They should end @ B+ minus maybe 10V or so as the CCS would need some headroom.


The CCS wont affect that impedance much. The plate load is there to set a load line for the tube, that's it. You need to look at the tubes internal resistance called Rp or Ra. The Rp stated in datasheets are for a specific operating point. It fluctuates some as well. There's a way to calculate the Rp of a specific operating point by looking at the grid curves. Remember, it's all about the curves!
There's also this formula that's useful.

µ = gm * Rp

Thanks @SonicTrance , yes the upper limit of the CCS load lines is too far, I drew them hastily so I could see the linearity lol with the CCS drawing 10V and a targeted B+ of 250-275 for my power tubes, it will only swing up to 240V or so.

Good to hear the CCS won't affect the impedance much, no reason not to use them then, we are moving forward listed Rp for the MH4 is 11kohm so I don't think it will be an issue, but I can calculate a more accurate Rp once the final bias point is chosen.

Just did some reading, I think I understand better now. The output impedance will be the combination of the Rp of the tube in parallel with the AC resistance of the load. For a CCS, the AC resistance is very high, on the order of 40Mohm or more. So taking those two values in parallel:

1/((1/11k)+(1/(40,000k)) = ~10.99k. Which is to say, the contribution of the CCS to the output impedance is basically zero. Your post popped up as I was writing this @A2029 , would you say this this accurate?

I need to find a crash course on LTSpice, I might take the time today. I'm a little bit wary of the accuracy of PSUD2 for modeling my power supply.

 

[L0rdGwyn]

By the way, thank you gentlemen for your input, nice to have others to bounce this stuff off of, Glenn gets enough PMs as is without me asking him all these questions

 

[leftside]

I think @A2029 also used LTSpice extensively when designing and building my amp.

 

[L0rdGwyn]

I think @A2029 also used LTSpice extensively when designing and building my amp.

I am going to have to bite the bullett and dive in, I will be happy I did later. As eager as I am to start building and testing, it would be better to take it slow and build good models. I have a laundry list of equipment to and parts to source anyway, it's gonna take some time.

 

[A2029]

@A2029 , would you say this this accurate?

With a CCS and a capacitor bypassed cathode resistor, Zout is approx equal to Rp at your chosen operating point.

I think @A2029 also used LTSpice extensively when designing and building my amp.

I did indeed, it's well worth learning to sim how the circuitry of an amp works and to get voltages and other parameters spot on.

 

[L0rdGwyn]

With a CCS and a capacitor bypassed cathode resistor, Zout is approx equal to Rp at your chosen operating point.

Perfect. I think this will be a very positive design change.

I did indeed, it's well worth learning to sim how the circuitry of an amp works and to get voltages and other parameters spot on.

Spot on, I like the sound of that. Well looks like my plans for the rest of the day are decided, better make a pot of coffee