Here is my Impasse build. I added pots to trim R5 to 8ma though the current sources. After testing, I couldn't get the THD at 4V peak unbalanced, 1kHz below about 0.05%. I removed the LEDs and added a 500R pots with a 300R resistors in series between cathodes and ground to adjust the grid to cathode bias voltage. After adjusting the grid bias I was able to reduce THD to 0.020% and 0.016% in the left and right channels respectively, much worse than what SY reported in his article.
I also occasionally see a low level oscillation at 42382Hz 60dB below one watt with lower yet harmonics at 42382Hz+/-120Hz.
The vacuum tubes are Electro-Harmonix 6SN7 and 6922 purchased from PartsExpress. There are probably better tubes for this preamp. Any recommendations would be welcome.
I also occasionally see a low level oscillation at 42382Hz 60dB below one watt with lower yet harmonics at 42382Hz+/-120Hz.
The vacuum tubes are Electro-Harmonix 6SN7 and 6922 purchased from PartsExpress. There are probably better tubes for this preamp. Any recommendations would be welcome.
I found that the THD would vary quite a bit with different makes of tubes.
Don't forget to bypass the resistors, the LED's have low a.c. Z.
Any particular tube that you would recommend that can be purchased?
Circuit analysis and SPICE simulations indicated that with cathode resistors on the order of 400 ohms, the difference in response with and without bypass caps was less than 0.1dB. I left them out.
Not to the level of expertise of SY -- but I plugged in several Raytheons and they were all over the lot.
The 6922 is not terribly critical since there's 100% feedback. The distortion is completely down to the 6SN7. I have only tried one sample of the EH and it actually was excellent, measuring similarly to the CV1988 I used in the prototype. Sample of one doesn't tell me much about variability!
Are you measuring distortion performance differentially? It is critical that both outputs be loaded identically- measuring one side only will not give you correct readings.
Are you measuring distortion performance differentially? It is critical that both outputs be loaded identically- measuring one side only will not give you correct readings.
Hi there, I notice that you have put a small heatsink on the CCS MOS-FETs, is this necessary ?
Calculations and simulations showed that the MOSFET dissipated about 0.8 watts, so I added the inexpensive heatsink. not being sure about the free-air dissipation characteristics of the TO-220 package.
I have made THD measurements of 2 kinds: differential with identical loads, and single-ended with the plate output shorted to ground. The measurements were about what was expected, with the single-ended THD about 1.4 times higher than the differential.
I figured out the low level 42382Hz oscillation. Florescent lights with electronic ballasts somehow coupled to the test lead of my Fluke multimeter. When I turn off the lights, the oscillation is gone. Surprising. At first I suspected a switching regulated power supply in a nearby computer or my Tektronics scope.
My hum measurements now look good: I measure about 0.5mV RMS noise on the cathode outputs to ground with the inputs shorted. In listening test, the hum is barely noticeable with the ear to the speakers.
I did some listening tests with cathode outputs into an F4 without the JFET input buffer. I have previously been driving the F4 using a BA3 front-end which sounds quite good. My initial impression of the ImPasse is that high frequencies are being corrupted by the input capacitance of the crippled F4 combined with the relatively high output impedance of the ImPasse. Spice simulations predicted this result.
Today I have made mods to my F4 to make it easy to quickly switch the F4 between crippled and non-crippled configurations. I will do some more listening tests with the F4 in the non-crippled state where the input capacitance will not be an issue.
Last edited:
More ImPasse listening tests:
I made mods to my F4 to quickly switch between crippled (no JFET input buffer) and uncrippled and conducted more listening tests.
The bottom line is that the Impasse sounds very good into the uncrippled F4, with only a couple of sonic situations that I am still evaluating.
Here is my setup:
Here are my (tentative) observations about various listening tests I have done:
I made mods to my F4 to quickly switch between crippled (no JFET input buffer) and uncrippled and conducted more listening tests.
The bottom line is that the Impasse sounds very good into the uncrippled F4, with only a couple of sonic situations that I am still evaluating.
Here is my setup:
- Buffalo-III DAC
- Preamp:
- none
- BA3 front end
- ImPasse
- DIY F4 amplifier using FQP19N20C and FQP12P20 mosfets. F4 modified optionally bypass the JFET input buffer.
- B&W CDM1 speakers
Here are my (tentative) observations about various listening tests I have done:
- BA3-FE into F4 with and without the JFET buffer: I have not detected any significant difference.
- ImPasse into F4 with and without the JFET buffer: Much better with the F4 JFET input buffer.
- DAC output direct to JFET buffered F4 vs. though BA3-FE: Obviously the signal levels are very low, but there wasn't and observed difference in sonic quality.
- STAX headphone listening: Much different listening experience, but the overall persection of distortion and frequency response was about the same.
I have the Stax Lambda as well, but not the nerve to use them bareback with the Imp. Do you have the Stax transformer?
I appear to be missing something. Why would I need an ImPasse with the STAX since the signal levels from the DAC are high enough to drive the STAX amplifier. I might understand some idea to drive the electrostatic headphones directly from ImPasse plate voltages. What am I missing?
lhquan,
Could you compare BA3+uncrippled F4 vs.Impasse+uncrippled F4 and which would you prefer?
Could you compare BA3+uncrippled F4 vs.Impasse+uncrippled F4 and which would you prefer?
Those two configurations both sound very good. I am still trying to understand (and repeat) a music passage which didn't sound quite right with the ImPasse. It involved transients (drum sticks against the edge of a drum set) that seemed to be more pronounced and perhaps distorted vs. the BA3-FE.
The bottom line is that I am still evaluating the sound of the ImPassse.