advertisement


Teddyreg Filter for higher voltage

C

cubeasic

pfm Member
Hello,
I have a question on how to use the teddyreg filter at higher voltages with LSK170 /LSJ74 FETs.
The application would be, for example, but not limited to the Hackernap frontend.

The LSK170 hasa 40V breakdown limit from gate to drain/source, the LSJ74 has a 25V limit.
My first excercices with LTspice show that at startup the gate voltage doesn´t rise fast enough to stay below the 25V (and 40V) limit of the FETs.
My thinking was to just use a Zener diode with e.g. 15V breakdown as shown here:

Teddyreg-Zener.png

The zener diode would maintain a maximum of 15 V across drain and gate during startup and else be practically invisible, because the d/g voltage is far below 10 V in regular operation.

LTspice says it might work as intended, but since I am not an electrical engineer and haven´t seen this anywhere I would like to have this sanity checked from folks that know what they are doing. Are there any drawbacks?

Thanks
Florian
 
Ah, I have not seen the Powerreg for a while. I think the Teddy Powerreg is intended for high current applications. It should have similar limitations then the regular Teddyreg because the FET gate charges through R3, R5 and R6, delaying the startup. I will have an input voltage between 45 and 60V.
 
Just thinking, do you need the 'accelerator' version as it was developed for applications such as clocks / DACs but isn't needed for power amps. Would this give more margin?
 
If you want to guarantee robust behaviour at higher voltage (and that includes peak current capability etc - consider possible in charging c6 and downstream decoupling to high voltages at start-up) ... why not just swap the 'fetlington' Q1-Q2 for a power Mosfet of suitable capacity instead? Cheap, and effective; probably cheaper than the jfet in fact.. and come with its own internal gate protection diode.

I'd probably re-route r7 to the output, picking a lowish value, as a discharge resistor to ensure the output discharges on shutdown, too.
 
... or Q1 could probably be replaced by a small signal mosfet or Darlington transistor... The current trough r7 helps to increase the transconductance of whatever device is used as Q1 and hence reduce the output resistance of the fetlington... Another "r7" at the output to waste some heat should further reduce output resistance whilst giving a discharge path for the caps as Martin said.
Voltage setting resistors will prob need changing to compensate for differing forward voltage of the output device/s with any of these methods.
 
Florian

I re-read your original post; a couple of other matters come to mind when using LM3x7s at higher voltages:

The 3-pin reg has a maximum withstand drop across it of about 37v. You can use LM3x7s at high voltage, so long as you ensure (esp in startup/shutdown) that the voltage drop across them, or reverse-bias on shutdown, is managed. You must do this for reliability above about 20v output anyway, and there are two steps that work IME:

1. Add another 1w/15V zener, connected between pins 3 and 2 on your digram, cathode at pin 2. It will avalanche at swtch-on to chartge the reg's output cap (c3) and so save the reg from excess forward voltage across it. This zener then also acts as the protection diode on shut-down, discharging the output side (c3!) back to the raw rail and limiting reverse-bias across the regulator to well under 1v NB use of this protection function diode even a basic diode is utterly mandatory with LM337s IME - they are fragile in this respect at turn-off.
2. Definitely also add the protection -diode shown in the LM3x7 datasheet from C2, for the same reasons.
 
martin clark said:
Florian

I re-read your original post; a couple of other matters come to mind when using LM3x7s at higher voltages:

The 3-pin reg has a maximum withstand drop across it of about 37v. You can use LM3x7s at high voltage, so long as you ensure (esp in startup/shutdown) that the voltage drop across them, or reverse-bias on shutdown, is managed. You must do this for reliability above about 20v output anyway, and there are two steps that work IME:

1. Add another 1w/15V zener, connected between pins 3 and 2 on your digram, cathode at pin 2. It will avalanche at swtch-on to chartge the reg's output cap (c3) and so save the reg from excess forward voltage across it. This zener then also acts as the protection diode on shut-down, discharging the output side (c3!) back to the raw rail and limiting reverse-bias across the regulator to well under 1v NB use of this protection function diode even a basic diode is utterly mandatory with LM337s IME - they are fragile in this respect at turn-off.
2. Definitely also add the protection -diode shown in the LM3x7 datasheet from C2, for the same reasons.
Click to expand...

Yes good point! Prob get away with it up to say 45-50V but well worth doing. I know from experience that it's mandatory at valve HT voltages:rolleyes:
 
Anything more than 10uf in either C3 or esp C2 and 337s are very susceptible to damage above about 25v in / 20v out. I've wasted hours and parts proving that for myself... ;)

[
and yep - I've a 317 delivering c.300Vdc reliably for my valve tuner, in use last 15yers sporadically, using only this approach.
The output needs to be calculated, tested and set to keep the total voltage drop across the 317 to a reasonable/ small figure, not least for heat management (remember the heatsink tab is at HV, so can't bolt it to a case...); but even at that a 15v or 24v or at most, 30v big fat Zener does an admirable job for pennies.

If anyone needs more than that - iirc Jim Williams wrote a nice Linear tech Appl Note on other approaches, good for up to 2kV ..!)
]
 
martin clark said:
... why not just swap the 'fetlington' Q1-Q2 for a power Mosfet of suitable capacity instead? Cheap, and effective; p
Click to expand...
This should be the Hackernap Version. I would like to benefit of the lower ouput impedance of the fetlington version. I could choose different FETs with higher voltage instead, but the LSK170 and LSJ74 are probably the lowest noise types one could find. I honestly don´t know if other FETs would make sucha large difference.
 
martin clark said:
I'd probably re-route r7 to the output, picking a lowish value, as a discharge resistor to ensure the output discharges on shutdown, too.
Click to expand...
Arkless Electronics said:
Another "r7" at the output to waste some heat should further reduce output resistance whilst giving a discharge path for the caps as Martin said.
Click to expand...
So another R7 -let´s call it R8 for clarity- would only see around 0,7V, the Vbe of the bipolar transistor. Which value schould I aim for here? Around 1k seems reasonable, adding another 0,7mA through the FET. But it could be much lower also.
 
martin clark said:
The 3-pin reg has a maximum withstand drop across it of about 37v. You can use LM3x7s at high voltage, so long as you ensure (esp in startup/shutdown) that the voltage drop across them, or reverse-bias on shutdown, is managed. You must do this for reliability above about 20v output anyway, and there are two steps that work IME:
1. Add another 1w/15V zener......
Click to expand...

I am aware of the LM317 limitations, and used the LM317 also for Hackernap-FE-duties with the protection diodes you mentioned.

From the LM317 cocooning I also got the idea of protecting the Fetlington-FET with a zener diode.
 
cubeasic said:
So another R7 -let´s call it R8 for clarity- would only see around 0,7V, the Vbe of the bipolar transistor. Which value schould I aim for here? Around 1k seems reasonable, adding another 0,7mA through the FET. But it could be much lower also.
Click to expand...

cubeasic said:
This should be the Hackernap Version. I would like to benefit of the lower ouput impedance of the fetlington version. I could choose different FETs with higher voltage instead, but the LSK170 and LSJ74 are probably the lowest noise types one could find. I honestly don´t know if other FETs would make sucha large difference.
Click to expand...

R7 sees the output voltage plus the Vbe of the bipolar. If loading is light then drawing more current by another resistor to ground at the output will lower Zout but there may be ample loading from the erm... load...

Unless you are wanting to power an MC phono stage or headamp etc the noise from a slightly noisier FET will be absolutely negligible. A Darlington such as MPSA14 etc should give even lower Zout if used instead of the FET.
 
You must log in or register to reply here.


advertisement


Back
Top