Borbely Fet Follower / HarryHaller?

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Disabled Account
Joined 2003
Prune said:
Can I simply put a 2SK389 in place of the IRF610? It seems to work in the Spice simulation, but I want to ask before I fry anything...


I am curious as to how this would have worked for the irf. those mosfets need to biased, unlike the jfets. Maybe you should post the schematic for us to see.

I think it is doable but unlike a drop-in replacement.
 
Disabled Account
Joined 2003
millwood, I'm not sure what you mean. How what would have worked for the IRF? I've been using the hexfets for some months already, driving my headphone amps. The DAC schematic I've linked to twice already in this thread. The LTSpice circuit of the output is here, with the IRF610 replaced with the 2SK389 (and a DC servo added, so I can ditch the large output caps).
I'm just not sure how to wire the volume control here.
 
Prune:

You won't need the DC servo circuit, since the '389s will give you zero DC offset if you bias them at Idss. You can do this easily by replacing R5/R7 with trimpots and adjusting manually until the output offset is 0. Or, replace D2 with a trimpot in parallel with a cap. R1 could be replaced with a CCS for better PSRR. Now just tie R8/R12 directly to ground. This will give you better performance than with the virtual grounds provided by the servo opamps, and you'll have a much simpler circuit. :)

Also, I'd think about increasing the value of R8/R12. 0.68uF and 1K gives you an LF roll-off of 234 Hz! You should be using something like 25-50K instead of 1K, and you can use slightly larger coupling caps like 2 or 4.7uF or so. I guarantee your 43122 will also sound better when it's not so heavily loaded.

If you want to add a volume control right in the DAC, the potentiometer goes in place of R8/R12.
 
Disabled Account
Joined 2003
Thanks for taking a look!
replacing R5/R7 with trimpots and adjusting manually until the output offset is 0. Or, replace D2 with a trimpot in parallel with a cap
The first one in simulation comes to about 11mA.
Which one of the two suggestions is better? Also, I'm worried about drift, otherwise I'd have just used a resistor divider for bias instead of a DC servo; do you not think drift will be a problem?
R1 could be replaced with a CCS for better PSRR
What would be a good CCS for this?
Also, would a CCS also help at the tops of the JFETS?
Now just tie R8/R12 directly to ground
Keeping the same value, 1M?
Also, I'd think about increasing the value of R8/R12. 0.68uF and 1K gives you an LF roll-off of 234 Hz
Huh? They are 1M, not 1K.

Let me ask this: can I skip the coupling caps completely and still adjust for 0 DC by a trimpot as you suggested (the common mode voltage at the DAC outputs is 2.75V)?

I was thinking of this type of volume control, as in the left half of the picture here:
http://www.diyaudio.com/forums/showthread.php?postid=343200#post343200
 
I've modified your LTSpice schematic to show what I'd do... I was just about to check the DC params, but realized I don't have a 2SK389 FET model! Can you email me the one you're using?

A good CCS for this application can be built using a 'ring-of-two' transistor circuit... I'll build a separate LTSpice schem. and email it to you.

1M? Oops, haha, yes I see that now. OK, that works fine, but a practical potentiometer will need to be a lower value. 50K or 100K would be fine in this position. I wouldn't go larger than a 100K pot, or you might experience HF roll-off with the input capacitance of the FETs.

Anyway, here's the modified circuit:
 

Attachments

  • buf_2.zip
    1.7 KB · Views: 135
Sure, and that's fine, but our intention here is to DC couple the outputs. Since the IRF610 version has cap coupling on it's output, the DC voltages don't matter. Also note that the IRF610's source voltage will always be lower than the gate voltage. If biased right, the source could in theory be at 0V while the gate sits at +2.75V. I haven't checked this, but I suspect that was the intention of jwb's circuit.

To get the JFET version to have 0V on the outputs, we would do well to have 0V on the gates, as long as we bias the JFET at Idss, or else we have either adjustable bias or self-adjusting bias like in the Borbely follower.

The volume control method you just referenced should work just fine, and unlike my proposal, won't require a 4-deck potentiometer, but just a single stereo pot. You should put the resistors and pot before the coupling caps, and then you can crank the value of R8/R12 back up to 500K or 1M.
 
Disabled Account
Joined 2003
I'm getting a few mV DC offset on the JFET buffer:
http://www.diyaudio.com/forums/attachment.php?s=&postid=343068
If I were to use a pot to adjust the lower resistance, what's the best way to wire it to minimize influence from the pot's parasitics on sound? In series with the lower 10 ohm resistor, or in parallel with it? In the second case ony a small fraction of the current will go through the pot as a large value pot will provide enough adjustment, but on the other hand the pot's capacitance is in parallel. What about a pot with resistor on top?
Biasing the input also works, but I can't use that as it messes up my volume control.
Can I expect the DC offset to be stable enough so that I won't need a servo?
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.