Introduction
Are you writing again about the “Slew” thing? There is a lot to be said about the slew rate still. In fact, this post is about some real life measurements and experiences had in the past few days.
This week I went to my friend’s (Tony) place to listen to his 01a with the additional follower discussed previously. Tony complained about that the addition of the follower changed the sound of the 01a. Something was lost, detail on the treble, airiness. I brought up again my test gear and surprisingly we found some H3 level which wasn’t there on the previous preamp. Odd, could that be the FET follower? In the end after several tests, I checked the operating point (you should never assume they should be as expected) and found that the follower wasn’t operating at the optimal level, source current was low and also the 01a was running out of place. I recalibrated the gyrator and voila. The sound of the 01a, came back!
Unfortunately we ran out of time and I did not take any measurements again after re-calibration.
The H3 bug
At home, I couldn’t stop thinking about the FET sonic impact. In the past I have used the FET as a follower to drive output stages, and it sounded really nice. The benefit introduced out-weighted any sonic signature, unfortunately I never had the chance to look at the harmonic plot of my implementations.
So, as usual, I couldn’t help myself and ended up building the following test rig:
Well, I didn’t use the same follower FET as Tony (which is one of my favourite enhancement mode ones I lately used), and instead I used DN2540. I took me 10min to wire this up on the bench and ran some interesting tests.
Testing the follower
I ran the tests at 2Vrms which is the maximum level I get on my sound card. I will re-run with the tests with the function oscillator to make some further comparisons if I get the time tomorrow. Distortion is really low (<90dB) @ 10kHz. H3 level is 5dB lower than H2:
Now, if we look at the performance of the follower at maximum frequency (i.e. 20kHz) we can see:
So the distortion here is higher at 0.01% (H2 @ -73.6dB). What happens if we reduce the CCS current down to 3mA (as the 01a will work on):
the distortion crept up to nearly 0.04% (H2=-62.6dB). Yes, here is the slew rate in full action!
Now, if we crank up the CCS current up to 20mA here is what you get:
The distortion goes down to 0.005% (H2=-80dB).
Running the follower at 10mA is ok, 20mA better if you can afford it!
I need to re-run these tests with 10Vpp again to see the H3 behaviour. Hopefully I find the time to do so as well as a comparison with other FETs
Hi Ale, you certainly have been busy. And getting interesting results. I experimented a little with FET SFs to drive the EL509 in my EAR859 amp in A2. I tried a variety of depletion FETS as I had them on hand; and its easy to cascode them. I preferred the IXYS over the DN2540. I then tried enhancement FETS as they seem to have higher transconductance which I think helped. I ended up adapting a Gary Pimm cascoded IRF820 CCS to SF duty. Perhaps that was “best” of all though I only used my untrustworthy ears for testing. best regards, tim
Hi Tim,
Good feedback. My experience was also better with higher GM enhancement mode FETs (e.g. STP3NK60ZFP) or other IXYS instead of the DN2540. In fact I’m not too precious of the DN2540, other than for CCS when voltages are not demanding.
I remember the DIYaudio thread you posted, good info there!
Ale
http://www.diyaudio.com/forums/tubes-valves/269883-source-follower-irf720-dn2540.html
tim
At these levels of distortion, I wonder if at some point noise becomes a bigger concern as you raise q current? Any thoughts on this Ale?
Thanks
Bob
Hi Bob, PS noise is always an issue. In fact, ground loops and any interference are far bigger worries than this. Having said this, in my experience if you achieve a quiet system with 10mA of quiescent current, not a big difference with 20mA, but supply design may require different components which create a constraint for sure.