diyAudio

diyAudio (http://www.diyaudio.com/forums/)
-   Tubes / Valves (http://www.diyaudio.com/forums/tubes-valves/)
-   -   Help with gyrator (http://www.diyaudio.com/forums/tubes-valves/199492-help-gyrator.html)

mogliaa 29th October 2011 02:17 PM

Help with gyrator
 
1 Attachment(s)
Hi all,
I'm in the process of optimising my 6J5 driver which currently has a CCS load.
Based on Wavebourn's circuit, I calculated the version to set the 6J5 at Va=260V and similar frequency response equivalent to a 150H ideal choke.

When I breadboarded the circuit, couldn't make it work. Although at some point I burned the LED array and got it biased properly by sheer luck. When I replaced back the LEDs, still couldn't make it work.

The PMOS is actually an FQP3P50, but couldn't find an LTSPICE model for it. I guess will have an impact on AC due to different capacitance, but should respond similar in DC? Haven't compared characteristics though....

The values I measured are in blue. Not sure about the value of the gate voltage as it's difficult to measure it with a voltmeter. When probe is connected it affects the impedance and then LEDs turn off until probe is removed - is this ok?

R11 is actually two 1M in parallel plus a 220K preset. R10 is two 22Meg in parallel.

The MJE350 emitter voltage should be at 324V, not the value highlighted.


Any suggestions where the problem could be?

Rod Coleman 29th October 2011 05:27 PM

Hi Ale,

The circuit (as drawn) can't work, I fear.

The LEDs will make the MJE circuit into a CCS only if the MJE350 can force about 324.6V on its emitter. With 316V on the emitter, the MJE350 will be degraded or destroyed, so you'll need a new one, sadly. See the specification for Veb(max), usually 6.0V.

Please post the Wavebourne original - I think the FET should be the other side of the MJE350 - to set the cascode voltage according to the stabilised voltage at the anode.

mogliaa 29th October 2011 07:26 PM

Hi Rod,
Yes, I thought the VCE measured might have killed the MJE350.
I think the circuit is ok, albeit my adaptation is not....see this thread:

http://www.diyaudio.com/forums/tubes...do-best-9.html

Thanks for the help
Ale

Rod Coleman 29th October 2011 08:08 PM

The most important omission is a resistor between the P-FET source, and the +ve supply. This sets the current for the triode's anode circuit.

The value is found from the voltage across R11 minus the Vgs of the P-FET, divided by the current you want to see in the anode. The voltage across R11 should be 10 - 20V for stable operation.

Next the cascode voltage needs to increase. This should be roughly (voltage across R11 + 10V), or 25 - 30V as a starting point. This will make sure the P-FET has enough headroom to stabilise the current.

mogliaa 29th October 2011 08:32 PM

Ok, so why simulation works perfect then? Where is the issue? Somewhere in the post is mentioned that you can omit the source resistor

Rod Coleman 29th October 2011 08:50 PM

If you omit the source resistor, the programmed current will be unstable, and always much larger than the triode can handle.

Concrete example:
If you adjust R11 to have 10V across it, and we assume the FET Vgs is 4V, then the source resistor must be 600 ohms to get 10mA, 300 Ohm for 20mA .. all the way up to 2000mA for 3 ohms! This will be independent of supply voltage. You can see that a source resistor of zero is not possible.

However, you will not get any current at all unless the MJE base is lower voltage than the FET gate - probably 10V lower, for stable results.

The simulation must work correctly even if the supply voltage is changed by 30V in either direction. If it does not, the circuit will not be stable in practice.

Alastair E 29th October 2011 10:26 PM

I cant see how that layout could possibly work anyway!

Where is the signal at the plate being developed and used as the input to the MOSFET?

--Its effectively removed by the 4.7uF cap!

Nah, I prefer a straightforward and highly effective single MOSFET, plate resistor, coupling-cap and pot. divider to set op.-point....

Super Simple, Great Performer....

Defiant 30th October 2011 05:07 AM

1 Attachment(s)
Quote:

Originally Posted by Alastair E (Post 2763307)
I cant see how that layout could possibly work anyway!

Where is the signal at the plate being developed and used as the input to the MOSFET?

--Its effectively removed by the 4.7uF cap!

Nah, I prefer a straightforward and highly effective single MOSFET, plate resistor, coupling-cap and pot. divider to set op.-point....

Super Simple, Great Performer....

Not that great. Any uncompensated semiconductor CCS is going to drift all over with temp :(

Try a setup like the one I attached. Rcs sets the current through the tube, while Rb sets the bias, and hence the quiescent voltage across the tube. D1 compensates against temp changes, thus keeping the current relatively stable over temp. Do NOT use the D45H11 - I just picked it for an example. It doesnt have the voltage rating for this circuit. Pick a transistor with Vce equivalent to your supply voltage and with an hfe as high as possible (but NOT a darlington!).

You notice that I used a resistor in the cathode circuit rather than LED's. The reason is that diodes are essentially a fixed bias, which will make the circuit sensitive to the characteristics of the tube. This is a Bad Thing™. You want the circuit to be as insensitive as possible to variations between individual tubes (and semiconductors). That is why a resistor is used to set the bias, as well as in the emitter circuit of Q1 to set the current (rather than using a divider in the base circuit). Also, depending on the characteristics of the diode and transistor there may not be enough voltage across D1 to forward bias Q1. If that happens, add some resistance between D1 and the base of Q1 . Do not add another diode or a transistor with the base tied to the emitter of Q1 like you see in a lot of SS circuits as it will screw up the temp balance. You can replace D1 with another transistor identical to Q2 with it's base tied to it's collector. In that case the emitter goes to B+ and the collector to the junction of Q1's base and R2. Provided both transistors come from the same wafer lot it will give you the best thermal tracking.

mogliaa 30th October 2011 09:32 AM

1 Attachment(s)
Quote:

Originally Posted by Rod Coleman (Post 2763237)
If you omit the source resistor, the programmed current will be unstable, and always much larger than the triode can handle.

Concrete example:
If you adjust R11 to have 10V across it, and we assume the FET Vgs is 4V, then the source resistor must be 600 ohms to get 10mA, 300 Ohm for 20mA .. all the way up to 2000mA for 3 ohms! This will be independent of supply voltage. You can see that a source resistor of zero is not possible.

However, you will not get any current at all unless the MJE base is lower voltage than the FET gate - probably 10V lower, for stable results.

The simulation must work correctly even if the supply voltage is changed by 30V in either direction. If it does not, the circuit will not be stable in practice.

Hi Rod,
Thanks for the help. Attached is an updated simulation. Have changed the LED array for a simple 33V zener to get the cascode voltage down to 300V - if I got it right from your suggestion above. Added the source resistor to set divider voltage as recommended and get the quiescent current to around 6mA. What it puzzles me now is that I need to reduce significantly the capacitor to 22nF to achieve the LP corner at 15Hz equivalent to the 150H inductor. From what I was reading in the thread mentioned before, the value should be higher that 1u at least.

What is driving the low frequency response then?

Thanks again for the help, I'm learning a lot with all this!

Cheers,
Ale

mogliaa 30th October 2011 09:34 AM

Quote:

Originally Posted by Alastair E (Post 2763307)
I cant see how that layout could possibly work anyway!

Where is the signal at the plate being developed and used as the input to the MOSFET?

--Its effectively removed by the 4.7uF cap!

Nah, I prefer a straightforward and highly effective single MOSFET, plate resistor, coupling-cap and pot. divider to set op.-point....

Super Simple, Great Performer....

Is this not the principle of the gyrator circuit? You want a fixed voltage in DC as would an ideal inductor provide, and then a high impedance response at high frequency.
What gyrator circuit do you recommend?

On a separate note, I tried multiple CCS configurations with excellent results!


All times are GMT. The time now is 02:59 AM.


vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2014 DragonByte Technologies Ltd.
Copyright 1999-2014 diyAudio


Content Relevant URLs by vBSEO 3.3.2