A while ago I asked Thomas how he would do a PP amp, and he mentioned differential with chokes in the common cathode, that would be at least one choke per channel. He shows 4, maybe he will connect both in series or parallel?
Well...I guess you want to do some maths to check...:
A 801a would need at least 15mA current (actually better 30mA), lets say he runs it at 400V. (PLease see his nice explanations here: VinylSavor: The search for that 'magical' operating point ) rp might be 4K if you are optimistic a bit (that is a result I see with many of those I have tested). 4K with an inductance of 100H for one full LL1668 gives us -1db at 25Hz or -3db at 6Hz. In SE.
340 Ohm winding resistance would give us a voltage drop of 340*0,015=5.1V. The bias you need at 400V is actually more like -35V at 15mA...all just for one tube.
With two chokes in series we would bust the defined limit of the chokes in the picture as this would mean 30mA over two chokes where each is speced only for 15mA in series with a voltage drop of approx. 20.4V.
With two chokes in parallel...this would reduce inductance to 50H effectively, which gives us still a -3db point of 12Hz...and he used not the 25mA, but the 15mA version of the LL1668, so maybe 150H/2 ?
A 801a would need at least 15mA current (actually better 30mA), lets say he runs it at 400V. (PLease see his nice explanations here: VinylSavor: The search for that 'magical' operating point ) rp might be 4K if you are optimistic a bit (that is a result I see with many of those I have tested). 4K with an inductance of 100H for one full LL1668 gives us -1db at 25Hz or -3db at 6Hz. In SE.
340 Ohm winding resistance would give us a voltage drop of 340*0,015=5.1V. The bias you need at 400V is actually more like -35V at 15mA...all just for one tube.
With two chokes in series we would bust the defined limit of the chokes in the picture as this would mean 30mA over two chokes where each is speced only for 15mA in series with a voltage drop of approx. 20.4V.
With two chokes in parallel...this would reduce inductance to 50H effectively, which gives us still a -3db point of 12Hz...and he used not the 25mA, but the 15mA version of the LL1668, so maybe 150H/2 ?
Really appreciate that thanks.
I'm going to ask a dumb question now, but just one of those weird things I need to get straight in my head.
Given that the valves are DHT, how best to configure the filaments ? Let's assume AC for now - simply run 5V AC across the 2 valves in parallel ?
I plan to use a pair of DHTs, perhaps 26s. It may not work well, but I'd like to try it.
Firstly, I've never used a CCS before (or any solid state) - is there a primer on how to calculate the resistor network required to provide 12mA through the CCS (each 26 approx 6mA) ?
Secondly, I've never used a valve model in LTSpice. If I PM you would you be happy to mail me that model to give me a start point ?
This is mich simpler: You built the ccs with a variable potentiometer, just order from kandkaudio the kits which are really inexpensive. Than connect plus and minus to a voltage source of 12V, can be any lab psu or battery or whatever you have. Dial the ccs with the poti to the current you want. Done.
Put it into your amp and it will maintain this current. Measure the voltage across plus and minus and be sure that it has enough headroom to operate (9-12V at least plus some swing for the signal) and that voltage is not too high for the potential heatsink you use.
What is really nice: You can plig and play with the bias point, eveluate the sound of more and less current, put other tubes in the circuit and bias will be automaticaly correct set.
For the filaments I use coleman regs.
Put it into your amp and it will maintain this current. Measure the voltage across plus and minus and be sure that it has enough headroom to operate (9-12V at least plus some swing for the signal) and that voltage is not too high for the potential heatsink you use.
What is really nice: You can plig and play with the bias point, eveluate the sound of more and less current, put other tubes in the circuit and bias will be automaticaly correct set.
For the filaments I use coleman regs.
If you want to have some real fun, grab some 4P1L (4П1Л) direct heated pentodes and triode strap them. They will have a mu of ~11 or so, but are an absolutely amazing valve, and still can be had for a decent cost. Only potential downside is the loctal basing, but they are such a physically and electrically handsome valve that it's definitely worth the fuss. They are often used in parallel SE for a wonderful little flea power rig, but make great signal tubes as well.
Just look at those triode curves!
Tube Tester Files - 4P1L
4П1Л/4P1L triode curves – Bartola(R) Valves
Just look at those triode curves!
Tube Tester Files - 4P1L
4П1Л/4P1L triode curves – Bartola(R) Valves
Attachments
Referring to the diagram in post #14 (page 2) of this thread, I'm assuming the trimmer resistor goes on the ADJ pin of the LM317 ?
The diagram shows a 27R on the OUT pin - how is that calculated ?
Sorry for potentially dumb questions, but I've never used any solid state before and just getting my head around it.
They are all over the place in the former soviet countries and plentiful on eBay as well. Lots of sellers in Russia, Ukraine, Hungary, etc. I'm not familiar withany specific UK based retailers but some folks will chime in hopefully to recommend some areas to check out.
RhythMick, look at the datasheet for the LM317.
http://www.ti.com/lit/ds/symlink/lm317.pdf
Page 12 gives and example of it used as a current source with the calculation.
I recommend reading the whole datasheet just for informational purposes.
Another nice DHT is the Type 49 with a Mu of 4.7 at 6ma when triode strapped. This tube was designed to be a dual purpose tube with Class B output characteristics in pentode mode, and Class A driver with the screen attached to the anode. However they are rather scarce and starting to get pricey.
http://www.ti.com/lit/ds/symlink/lm317.pdf
Page 12 gives and example of it used as a current source with the calculation.
I recommend reading the whole datasheet just for informational purposes.
Another nice DHT is the Type 49 with a Mu of 4.7 at 6ma when triode strapped. This tube was designed to be a dual purpose tube with Class B output characteristics in pentode mode, and Class A driver with the screen attached to the anode. However they are rather scarce and starting to get pricey.
Last edited:
I built Pete Milletts low mu preamp and i love it. You can change tubes to increase the output level. Here is a link to his page and a pic of mine.
Low-mu preamp
Low-mu preamp
Attachments
I could never successfully reconcile all the conflicting requirements for a diff stage. As a consequence my balanced pre is not balanced at all but still works and sounds well, thank you 
Why conflicting? Only because of severe compromises as far as perceived sound is concerned. At least perceived by myself.
Issue No 1: The need for a negative PS. With all the tremendous effort many of us put towards B+, the negative supply should be of similar quality, right? So, uhm, another mercury rectifier, two more amorhous chokes...becomes a bit excessive, doesn't it?
Issue No 2: Sand in the tail
A simple jfet may actually work as a modest ccs within the confines of negative grid bias. But does it sound any good? Not at all.
Something i have not seen done is to use a "proper" ccs from B+ and a current mirror in the tails. The current mirror needs a volt or two and should also fit within the grid bias voltage. Would it sound better than the jfet? With all the additional sand?
The choke CCS idea certainly grabs me. How does Thomas rectify the tail supply?
Why conflicting? Only because of severe compromises as far as perceived sound is concerned. At least perceived by myself.
Issue No 1: The need for a negative PS. With all the tremendous effort many of us put towards B+, the negative supply should be of similar quality, right? So, uhm, another mercury rectifier, two more amorhous chokes...becomes a bit excessive, doesn't it?
Issue No 2: Sand in the tail
A simple jfet may actually work as a modest ccs within the confines of negative grid bias. But does it sound any good? Not at all.
Something i have not seen done is to use a "proper" ccs from B+ and a current mirror in the tails. The current mirror needs a volt or two and should also fit within the grid bias voltage. Would it sound better than the jfet? With all the additional sand?
The choke CCS idea certainly grabs me. How does Thomas rectify the tail supply?
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.