That is one road to follow, hope that NFB will cure all ills , or
take the other road and
try like hell to make your circuit as linear as possible before
NFB.
Since this is DIY and the constraints of commercialism are
less , why not make it better ?? (more linear)
perhaps as good as a double blind test is to A/B 2 amps
while listening to thier OP's with a pair of headphones
(Be careful !!).
I recently did this with my simple self amp and my new one
(which does the .0004%/1W thing)(attached). I could hear a better
soundfield with the new one (headphones), but little difference
when run on typical loudspeakers. I think GK just takes the
"high road" in his designs , as he also derives educational
benefits from his pursuit of excellence. 🙂
OS
Attachments
G.Kleinschmidt said:
No need to guarantee the uniformity. Regardless the distribution of NPN vs PNP spreads, the distortion will never be higher. Even in the extreme case when one of them is zero and the other enormous, the distortion is still slightly less.
I think Glen was referring to thermal runaway and assorted issues, not THD per se.
Jan Didden
Jan Didden
janneman said:
Hi Jan,
He was addressing two issues:
1. Indeed, thermal runaway and/or current hogging.
2. A low bias voltage on only ~13mV, see:
http://www.diyaudio.com/forums/showthread.php?postid=1757356#post1757356
It is the latter to which I was referring. IF the amp is thermally stable with Vbias=26mV (and why not? D. Self has done it too), then switching to a lower bias will not make the amp less stable.
edit: Because this is rather obvious, I was wondering if there were other consequences like more THD, for example.
Regards,
Edmond.
PMA said:
This being audio with a very high crest factor, peak is the most
appropriate figure, and is clearly identified in any case.
😎
Wow, I must have real problems with this design!
Actually, sometimes I have had a few problems. In one case, I just re-beta matched the output devices, and it worked OK. In another case, we redesigned the heatsink internally to better balance the heat distribution. We had prototypes built, before we went into production. This is where REAL PRODUCT is better and easier to test, than a SIMULATION.
Actually, sometimes I have had a few problems. In one case, I just re-beta matched the output devices, and it worked OK. In another case, we redesigned the heatsink internally to better balance the heat distribution. We had prototypes built, before we went into production. This is where REAL PRODUCT is better and easier to test, than a SIMULATION.
John has regular global negative feedback in his JC-1 (not weak), so the 'problems' are reduced 😉
PMA said:
To my experience, and after having the opportunity to measure and listen a pair, the JC-1 is certainly a very good amp, excelling in muscle and ability to handle difficult loads. It is rock solid, very well build and, given the performance and specs, certainly sells for a decent price.
The measured performance is mediocre, however I was not expect anything more for the money. The JC-1 certainly confirms that this level of objective performance is anything that a customer, more or less experienced in the high end world, may look for.
I do not have any idea on if/when/how the customers are using the bias switch, but certainly while listening to the JC-1 I felt no urge to touch it, at all power levels (don't think I went over 200W/channel).
andy_c said:
Hi Andy,
This is a very good explanation.
It is worth pointing out that under all of these conditions, spectral analysis of the distortion can be important.
Moreover, use of a spectrum analyzer can eliminate the N in THD+N so that one does not have to "assume" that rising THD+N at low levels is always just noise.
If by spectral analysis all orders of distortion are less than -100 dB at all output levels, and for all fundamental frequencies up to 20 kHz, it is very difficult for there to remain any nonlinearities that are audible under any conditions.
Cheers,
Bob
Terry Demol said:
I don't suppose circuit analysis or analogue design is your forte.
But it is a DIY forum so I take it that comments from people like Scott Wurcer don't qualify.
The JC-1 does have too much 2'nd harmonic distortion. I would like to lower it and get more impressive specs.
Bob Cordell said:
Yes, of course - agree completely.
Maybe this is nitpicking, but if one removes the noise from the measurement by spectral analysis, I'd call the result "THD" rather than "THD+N", and then assume that rising THD at low levels really is due to crossover distortion. If there were a plot of THD (broadband noise removed by spectral analysis) vs. level, it would be good to see a dotted line where the noise floor of the measurement is reached, if applicable.
If the crossover distortion is bad enough, then rising THD+N (that is, without spectral analysis) at low levels could be crossover distortion-related, as Nelson mentioned in his online article a while back.
I agree with that 🙂.
Nelson Pass said:
Hi Nelson,
I was thinking of this statement:
This is on this page of the six moons article, just above Figure 6.
Edit: You didn't mention THD+N specifically though.
Edit 2: Also, we talked about this before here
- Status
- Not open for further replies.