[Pars]

Back to the coupling caps for the SA5.1...

I sent an email to the designer today:
I am about to replace the stock line stage coupling caps and wanted to use 1uf Dynamicaps in order to retain the shields if possible. I was a bit concerned with the frequency response issues posed by R17 of 100K ohms. Would you recommend increasing the value of R17 or is a -3dB point of 1.6Hz normally acceptable? Or do you think that using an 8uf cap here is a better solution

Reply:
The SA-5's line stage is a two-pole design with input and output coupling capacitors. Proper selection of capacitors needs to be done by loading the SA-5 into the power amplifier's input impedance (or resistor of same value) and then adjusting the poles to achieve as long a bass step as possible with as little overshoot as possible.

Goes on to explain that explaining what "bass steps" are and how to measure them is beyond the scope of this "freebie" email... OK

I've never heard of the term bass steps, but seems like a bit of BS to me? Any thoughts or illumination appreciated.

 

[luvdunhill]

Quote:

Originally Posted by Pars /img/forum/go_quote.gif Back to the coupling caps for the SA5.1... I sent an email to the designer today: I am about to replace the stock line stage coupling caps and wanted to use 1uf Dynamicaps in order to retain the shields if possible. I was a bit concerned with the frequency response issues posed by R17 of 100K ohms. Would you recommend increasing the value of R17 or is a -3dB point of 1.6Hz normally acceptable? Or do you think that using an 8uf cap here is a better solution Reply: The SA-5's line stage is a two-pole design with input and output coupling capacitors. Proper selection of capacitors needs to be done by loading the SA-5 into the power amplifier's input impedance (or resistor of same value) and then adjusting the poles to achieve as long a bass step as possible with as little overshoot as possible. Goes on to explain that explaining what "bass steps" are and how to measure them is beyond the scope of this "freebie" email... OK I've never heard of the term bass steps, but seems like a bit of BS to me? Any thoughts or illumination appreciated.

sounds like BS to me... you mean that you have to adjust xover poles in the preamp to a specific power amplifier? how ridiculous is that...

 

[nikongod]

Quote:

Originally Posted by Pars /img/forum/go_quote.gif Goes on to explain that explaining what "bass steps" are and how to measure them is beyond the scope of this "freebie" email... OK I've never heard of the term bass steps, but seems like a bit of BS to me? Any thoughts or illumination appreciated.

as a scientific wild ass guess, that may mean that they tune the coupling cap size with a low-frequency square wave.

my mildly useless, totally speculative drivel to follow.

Obviously a larger coupling cap will have a lower -3db point, but they may have the idea that using a cap which is too large brings its own problems to the system. I dont know if there is any factual basis for this idea or not. With an oscilloscope and square-wave generator you could see whether this has any merit or not in like 5 minutes.

after writing that, I pondered why they would not use it to tout the technical superiority of their products were it the case. to add to that, it may not *really* depend on the fact that the input impedance of the amp being driven is within a certain range, but that the actual size of the cap does not exceed a certain value.

 

[Pars]

You have to remember that he is now basically modding or servicing his old equipment, plus has his new company Aria. Seems as if he has become more defensive when dealing with DIYers (such as ermm, me) and protective of his revenue stream, so he isn't going to give me a clear answer. And I completely understand that, not dissing him at all. I had talked to him 2-3 times in the past and he had been pretty approachable, so I figured it was worth a shot.

Thanks for your input (Marc and nikon).

 

[Pars]

Quote:

Originally Posted by dsavitsk /img/forum/go_quote.gif <snip> Not quite, but the datasheet says that 1M is max, so 500K is certainly fine. This resistor just connects the grid to ground usually (here to the bias supply, but ground as far as the tube is concerned) to reference it. There isn't any current flowing through it. By the way, a CCS on the plate would be a nice improvement.

Regarding a CCS, something along these lines? It appears that the board run for this did not materialize? Nice answer to the one insistent poster btw

Is there a particular one of the 6 circuits (if I counted right) that would be the preferred one for my application? I was thinking about using the IXYS FETs (I think I have 1 or 2 laying around, the 10m45 is it?).

 

[regal]

Yes the IXYS 10m45, I am actually modifying a DAC to use these for a CCS, it is recomended to use 2 as a cascode on the plate. You can set the current using a 9V battery and a pot before you hook it up.

 

[dsavitsk]

Quote:

Originally Posted by Pars /img/forum/go_quote.gif It appears that the board run for this did not materialize? Nice answer to the one insistent poster btw

I actually ran the group buy that did happen which was for the IXYS boards. The boards can be seen in regal's post above. (Edit: actually, those aren't the ones that were used as the design as revised a bunch.) The answer I gave to b.f. was a hint for him to send me an email as I had some extra boards I could have supplied him with. But, he was so insistent on getting angry that I guess he didn't get it. Oh well. The boards ultimately were sold off to someone else.

Quote:

Is there a particular one of the 6 circuits (if I counted right) that would be the preferred one for my application? I was thinking about using the IXYS FETs (I think I have 1 or 2 laying around, the 10m45 is it?).

I think the one shown above is a good option. But, even a single IXYS chip is a big improvement from a resistor and worth experimenting with first.

 

[regal]

These are easy to build on a RadioShack perf board. I am fitting 2 casscodes for the plate and 2 singles for the sinks on a single little clad ratshack board.


Question I have is when using the CCS on the cathode as a sink do you have any improvement on PSRR vs a resistor?

 

[Pars]

I etch alot of my own boards, so I could start with that. Do the IXYS parts require heatsinking? The size of those look like they could be implemented rather easily as the current resistors are Mills MRA-12s (R20, schematic in post 41, page 5). Possibly stand these boards on edge and drop them right in. I assume I would replace R20 with a CCS? Or would it be better to replace each of the 51 ohm resistors with a CCS?

 

[dsavitsk]

Quote:

Originally Posted by Pars /img/forum/go_quote.gif Do the IXYS parts require heatsinking?

Not always, but often. It is pretty easy to figure out how much heat is dissipated. As an example, a CCS loaded 6dj8 with 10mA of current and biased to 2V will have about 100V on the plate. With a 200V supply, this means dropping ~100V across the CCS at 10mA. W = VI = 100 * 0.01 = 1W. You need a sink for that. Lower voltage, or lower current, you might not. For the version with 2 IXYS chips, it is a little more complicated, but most of the heat is dissipated across the top chip.

 

[regal]

Where are getting 1W=heatsink from the data sheet? I am looking at 0.5W and wasn't planning on using a sink, what's the limit?

 

[luvdunhill]

Quote:

Originally Posted by dsavitsk /img/forum/go_quote.gif Not always, but often. It is pretty easy to figure out how much heat is dissipated. As an example, a CCS loaded 6dj8 with 10mA of current and biased to 2V will have about 100V on the plate. With a 200V supply, this means dropping ~100V across the CCS at 10mA. W = VI = 100 * 0.01 = 1W. You need a sink for that. Lower voltage, or lower current, you might not. For the version with 2 IXYS chips, it is a little more complicated, but most of the heat is dissipated across the top chip.

I didn't realize that the CCS would drop voltage in this case. I suppose just sourcing 10mA of current without a voltage drop would perhaps violate some fundamental law, but I assumed this is how these 3 legged devils worked. Can you explain?

 

[Pars]

Quote:

Originally Posted by luvdunhill /img/forum/go_quote.gif I didn't realize that the CCS would drop voltage in this case. I suppose just sourcing 10mA of current without a voltage drop would perhaps violate some fundamental law, but I assumed this is how these 3 legged devils worked. Can you explain?

In my case, between R20 (10K ohm) and the 51 ohm resistor(s), B+ is ~250Vdc and the V at the tube is ~70V, so it is dropping ~180Vdc. The current is therefore ~18mA. The IXYS device has to have some effective resistance, and any current thru it is going to result in a voltage drop across it (Ohms law). Add in RTEST (or is this jumpered?) and R3 and the pot || R2.

 

[regal]

Guys the voltage drop is determined by the tube. You are setting current fixed via the IXYS and cathode bias via your cathode resistor, so the tube pushes its plate voltage to match. Based on the plate curve. Its backwards of how we were originally taught biasing tubes but this is how it works.

 

[dsavitsk]

Quote:

Originally Posted by regal /img/forum/go_quote.gif Guys the voltage drop is determined by the tube. You are setting current fixed via the IXYS and cathode bias via your cathode resistor, so the tube pushes its plate voltage to match. Based on the plate curve. Its backwards of how we were originally taught biasing tubes but this is how it works.

Yes. You can read my more thorough description at my MEHA Page.