John Curl's Blowtorch preamplifier part III

Status
Not open for further replies.
Howie, I had a hand in the original studio and transmitter site install and subsequent tech duties for this community fm service.
The studio was moved a few years ago and a different final processor/multiplexer installed.
AFAIK the same analog STL is in use and the streaming feed is from an off-air reciever -
RRR-01

Still sounds quite good to me, certainly way more listenable than the local commercial FM offerings Listen Live/Podcast Playback | 89.7FM Perth.

I'm interested to hear your 320k streaming feed.


Dan.
 
Well, your boss says "Let-us study a new range of products for the next hifi show."
So, you begin to study with your colleagues this 150 watt amplifier of your dreams. With the ideas you prefer at this moment, and those of your collègues, your habits, new ideas you mostly had during week-ends, some you want to experiment or had patented, the components on the shelves, and some new components you can find near the providers of the company. Some brain storms, calculations, simulations, designs, improvements, listening, improvements etc... You know how it works...
And you finish with a nice prototype.
Then your boss comes and ask for compromises, less expensive components etc.. And you try to save the most you can... the result depends of you authority, your dialectic, your talent to convince...
And, at the end, you finish with a prototype witch is, let says, 50% of what you wanted. If you are good enough, the essential is saved.

At this moment, the marketing department ask-you for beautiful stories that will help them to sell this future product. They will understand some of your explanations (or not), and will, they are paid for this, invent fairy tails about it ...
Some will be kind of snake oil, what the hell. Are-you responsible of this Bull sh.. ?
I can remember very funny moments, reading the literature of those guys about my own work..
At the end, If the result is good enough, despite all that, for you can enjoy music when this amplifier will reproduce-it, if you are satisfied enough with-it to get one prototype in your home for your own system...
Job done.
The truth ? Where is *the truth* in this process ?

Alone in his home, with the freedom he has, the same guy working in a DIY process has all the luck to get better results.
(Please, be indulgent to my poor English ;-)
 
Last edited:
I too miss SY around here. His absence means more work for people like Scott. In his last few months on here I think I detected an increasing short-temperedness with people talking nonsense, so maybe he had to quit for the sake of his blood pressure? I am sometimes tempted to take the same action, as repeatedly teaching basic electronics to reluctant students (including those making money from audio) can become tiring.

It does get frustrating that you wait a few weeks and the same stuff just comes back. I'm going to try and extract a hypothesis and a verifiable FULLY parameterized experiment before making any comments first and see if that helps.
 
I suspect you will see low frequency components differ with capacitor type.

So DA introduces an error component that is essentially linear. Big day today. John and Scott agree.

Only when there is non-linearity i.e. C and or R are not constants at a specified level.

Could you please specify a complete experiment, schematic and levels for the two tone test above.
 
Tournesol, you have it right! That is exactly how I operate.
Even today, I had to talk to my boss at Parasound about some 'feature' he wanted to add to a new amp. I reminded him of the difficulty in doing it without seriously compromising the design, and we instead went with an alternate solution. Last week, he was in Taiwan at the factory, and was pressured to change the output devices, the number of output devices, and the operating temperature of my new designs. We both stuck to our principles, against the 'practical' engineers who actually manufacture Parasound's amps.
Just today, I got a private message from a diyer here about biasing my power amps. Somebody, somewhere wants to UNDER-BIAS my amp designs. Why? Because it will still meet specs and run relatively cool, but it WILL have more Xover distortion. I just try to make the best that I can that the company can manage to sell. I have NO input into marketing, and they know better than to try to get me to help them above what I always talk about here. In fact, they don't like me to talk to reviewers about my designs, because I might undermine their marketing efforts. I would not lie just to promote a product. What is the point in that? No matter what some here imply.
 
Member
Joined 2014
Paid Member
Well, your boss says "Let-us study a new range of products for the next hifi show."
So, you begin to study with your colleagues this 150 watt amplifier of your dreams.

Outside of the Fred in a shed one man boutique audio companies (many of whom go bust) the initial design brief usually includes a selling price. Up front you know what your BOM cost is and cry that the metalwork will cost 3x your component budget. Companies who make statement products rarely intend to sell more that a small handful. A few lucky ones who have the name can sell quite a few, but they can be counted on the fingers of one hand.
 
Disabled Account
Joined 2012
OK.

So, we said (we?) said that some non-ideal behaviour of electrolytic caps can have an affect on the amp and/or sound just thru dc leakage. Undesirable, Indirect effects (line #131);

What about another non-ideal characteristic ... changes in series Z of the cap with frequency? (used in fb - port to ground thru Rg.) Hmmm. Lets see what people actually use. Using 1K and 100mfd; -3db at 0.7Hz. Should be low enough. We need to have less than 0.1dB amplitude difference to be undetectable in listening. The guys at Harmon found we could detect 0.1 dB change with low Q with music.

So, we need capacitor series Z, Xc+ESR funnies to be under approx 1% or under 10 Ohms variation. using a best case -- Non-Polar (NP) electrolytic..... testing shows 66 Ohms at 20Hz. Oops. NFG. 9.5 Ohms is at 150Hz. A 220mfd is better at 40 ohms/20Hz. But, in both cases --- 100 and 220mfd... what's happened to that nice ideal 90 degree capacitor? Its down to 18-13 degrees at 20KHz.

DSC03114.JPG 20Hz/100mfd



DSC03119.JPG 20KHz/220mfd



Lets go way up to 2200mfd;

DSC03120.JPG 20Hz/2200mfd


Z is low enough, but now the self resonance of the large value cap has gone inductive within the audio BW.


DSC03122.JPG 20KHz/2200mfd (went inductive)


How does either of these caps in the fb affect the error correction --- varying degree -- you compute how the fb is affected.

Can it be detected by listening to either of these against no cap at all? With no Z or sever phase changes -- only R's.
And, the DA of the 2200 being 5-15% absorbed in Cda is significant amount to be released later at a longer TConstant.


A dc-servo is beginning to look better and better.



THx-RNMarsh
 
Last edited:
Administrator
Joined 2004
Paid Member
Hi Richard,
I really like your meter!

One day ... because one of the secrets of great design is knowing your components. By this, I mean, to really know your components on the bench.

DC servos are only going to look better if they are implemented in the right place. Given the choice, I`ll happily put up with a little DC offset instead of a servo.

-Chris
 
Disabled Account
Joined 2012
Hi Richard,
I really like your meter!

One day ... because one of the secrets of great design is knowing your components. By this, I mean, to really know your components on the bench.

DC servos are only going to look better if they are implemented in the right place. Given the choice, I`ll happily put up with a little DC offset instead of a servo.

-Chris

If you can get away without it, good way also. I dont use one on my little HPA which is direct-coupled.


-Richard
 
AX tech editor
Joined 2002
Paid Member
OK.

So, we said (we?) said that some non-ideal behaviour of electrolytic caps can have an affect on the amp and/or sound just thru dc leakage. Undesirable, Indirect effects (line #131); <snip>
THx-RNMarsh

But these are frequency response deviations right? Nothing non-linear?

BTW I assume that your values are meant to be uF, not mf (m normally is milli).

Jan
 
Richard, thanks for your latest measurements with your impedance meter. I have nothing like that, so I have been further enlightened as to what is wrong with AL caps.
(Also, Richard, we old guys sometimes write "uuf" instead of ''pf'' and I like ''cps'' sometimes, but dare I use it? as Jan will correct me '-)
 
AX tech editor
Joined 2002
Paid Member
Richard, thanks for your latest measurements with your impedance meter. I have nothing like that, so I have been further enlightened as to what is wrong with AL caps.
(Also, Richard, we old guys sometimes write "uuf" instead of ''pf'' and I like ''cps'' sometimes, but dare I use it? as Jan will correct me '-)

I am aware of that John, but I think we all benefit if we speak the same language. There IS a difference between mΩ and MΩ, between mF and uF.
There is also the small issue of respect for your partner in being as precise as you can...

Edit: taking the risk to be pedantic, Farad is a name so needs to be capitalized. Like Hertz (Hz), Tesla, Henry. 5uF, 10uT, 2GHz, 20mHz, 10MHz, 30mH. Easy and consistent when you know it. ;-)
I'll take off my editor's hat now.

Jan
 
Last edited:
Disabled Account
Joined 2012
One is an amplitude error, the other is related to minimizing amplifier thd and IM via fb phase shift from 90-0 degrees. And, maybe even stability. Somethings not seen in CAD modeling of the amp using perfect caps with no 90 degree change vs freq; no leakage currents flowing, no DA, etc. VS no amplitude error and no phase shift thru fb cap. Subtle but detectable compared to dc-coupled/no capacitor.

The value needed just for the amplitude to be << 0.1dB is a lot larger than most designers use..... they calc just for -3dB point and assume it is going to be fine.

Really large C values are big in size and more expensive. Larger has the phase go to zero sooner (lower resonance freq).

Yes, microFarad.


I dont know where this obsession with linear and non-linear comes from. There are plenty of errors from the large parasitic elements of electrolytic caps to interfere with making the amp sound its best. Not developing voltage across the cap is only one.

Film caps have same parasitic elements but they are so smaller as to make much less difference in a comparison. But, then they are large and cost many times more and so not used in mass produced amps/recvrs.

This is just with a low ac voltage and low current. Other issues arise in speaker cross-over or a PA with cap coupled output. The large Amperes of currents put a premium on the esr being stable (it isnt) and lead attachment method and materials..... all are parts of esr including the dielectric heating. Which in turn changes the phase angle further. Internal heating affects just about every parameter then. Changing away from spkr bipolars to film - especially film and foil do a Lot better.

These are a few of the reasons why changing a polar or bi-polar to film or dc-servo makes a detectable difference. The details are in the parasitics.


.
THx-RNMarsh
 
Last edited:
AX tech editor
Joined 2002
Paid Member
One is an amplitude error, the other is related to minimizing amplifier thd and IM via fb phase shift from 90-0 degrees. And, maybe even stability. Somethings not seen in CAD modeling of the amp using perfect caps with no 90 degree change vs freq; no leakage currents flowing, no DA, etc. VS no amplitude error and no phase shift thru fb cap. Subtle but detectable compared to dc-coupled/no capacitor.
.
THx-RNMarsh

Dick I don't think that is correct. The ESR (or ESL for that matter) is in series with the R in the lower feedback arm, thus can be lumped in with it.

Instead of 1k with the ESR-less cap you now have, say, 1.01k. Slight change in freq response, and slight change in THD because the feedback factor changed a fraction of a dB. Much less than the usual 5% or 10% of the cap variation, which has exactly the same effects.

Jan
 
AX tech editor
Joined 2002
Paid Member
I dont know where this obsession with linear and non-linear comes from. There are plenty of errors from the large parasitic elements of electrolytic caps to interfere with making the amp sound its best.
.
THx-RNMarsh

No obsession I think, but one causes distortion and the other a change in freq response, tonality if you want. Very different.

Jan
 
AX tech editor
Joined 2002
Paid Member
These are a few of the reasons why changing a polar or bi-polar to film or dc-servo makes a detectable difference. The details are in the parasitics.

.
THx-RNMarsh

Yes, agreed. As I explained above, servos may be orders of magnitude more non-linear than a reasonable good cap.

Edit: for some reason I repeated a post above. Apologies.

Jan
 
Status
Not open for further replies.