Jung current source article in latest audioXpress

[bcarso]

There is a lengthy and authoritative piece, and only part one at that, by Walt about current sources in the latest (April 2007) audioXpress. The focus is on PS rejection and output Z, among other parameters. I recommend purchasing an issue and supporting this worthy if often uneven magazine.

Parenthetically, a sidebar comment from Bruce Hofer about Ap systems reminded me of what Marshall Buck recently told me, that Ap is starting to qualify their degree of support for older systems, citing the difficulty of getting parts. A sad day in a way, but then how much stock can they be expected to carry for all these years.

 

[FredForssell]

I totally agree with bcarso. Just when I think I might let my subscription to the mag expire, an article like Walt Jung's comes along. This is a great piece of work by Mr Jung on an oft overlooked area of circuit design, and it is articles like this one that make Mr Dell mag valuable to me.

As for AP... yeah I often wonder about certain aspects of that company, but I never wonder about the test gear that they design. Top notch stuff. I upgraded from System One to a System Two Cascade about 4 years ago and I'm not sure how long they'll support even that model. None the less, it is really, really, good equipment, imo.

Cheers

 

[NewYorkDave]

Recently, after campaigning for it for six years, my employer bought me a Portable One +. I love it!

 

[Samuel Groner]

And if you search hard enough: www.audioxpress.com/magsdirx/ax/addenda/index.htm

Looks like a worthy article indeed--thanks for the hint!

Samuel

 

[bcarso]

[quote author="mediatechnology"]Sam - Thanks for the post and link. Did you see the low noise measurement preamp cite below Jung's?[/quote]

Of course you will still go and buy the issue, right? :razz:

That preamp article is in the same issue, as you may have noticed. I'm happy to see that he is using some of the Linear Integrated Systems parts, although I wonder a bit at using a dual with the sections in parallel---although you do improve noise performance by 3dB, one could as easily use a couple of fairly well-matched SK170's. Perhaps the 389 was what he had around, and there's the appeal of having no need to look at matching.

 

[clintrubber]

[quote author="bcarso"]Parenthetically, a sidebar comment from Bruce Hofer about Ap systems reminded me of what Marshall Buck recently told me, that Ap is starting to qualify their degree of support for older systems, citing the difficulty of getting parts. A sad day in a way, but then how much stock can they be expected to carry for all these years.[/quote]
Hi Brad,

Is there any info on which models this applies to ? Or was it a more general statement ?

I had a brief look at their site but couldn't find anything yet.

Time to get the older AP-equipment at work in good shape before it's sold to employees :twisted:

But serious, we may give them a good look and get any repairs done while still doable.

Thanks,

Peter

 

[clintrubber]

[quote author="mediatechnology"]Peter;

AP sent me an e-mail with a table. I'll see if I can find it.

Nope think I deleted it but it's in the recent audio.tst newsletter. You should be able to find it on their site.

Wayne[/quote]

Hi Wayne,

Thanks, I have it, as it happens just from a few days ago. It was also (still) in my Inbox, hadn't read that mail fully yet.

Thanks :thumb:

Peter

 

[clintrubber]

[quote author="mediatechnology"]Peter - Maybe you can also find my lost copy.

Since I'm such an idiot and went ahead and deleted mine you might want to take a screenshot of the table and post it in the lab or brewery. [/quote]

No problem.

Excuse me for being lazy for now and pasting it here (after all, AP was already brought up in the first post of this thread :wink: )

Here we go (& please ignore the formatting-mess):

Sound Advice: Audio Test Q&A

Q: What is the AP policy regarding support for System One and other discontinued models?

A: Audio Precision products that have been discontinued for a period of at least five years are placed in a product life-cycle category called "Conditional Support". Conditional Support means that Audio Precision can no longer guarantee its ability to service or repair a product due to the unavailability of certain key components or assemblies, or depleted repair stock inventories.
Audio Precision will continue to provide hardware service for these older products on a case-by-case basis subject to the following procedure:
The customer must first contact our service department for an initial consultation, then download and run an approved Performance Check that will assist us in evaluating the condition of the instrument.
If AP can determine the most probable failure(s) from the results of the Performance Check AND determine that all necessary repair parts or assemblies are still available, the customer will be instructed to fill out a Service Request form and send the product to AP's US factory or to a specified service center. At that time, a non-refundable evaluation fee of $250 will be due, and AP will provide a non-binding estimate of total service charges. If AP cannot make a tentative diagnosis from the Performance Check results, or if it is totally non-functional, the unit can still be submitted for review subject to the non-refundable $250 evaluation fee.

Successfully repaired units will have the $250 evaluation fee credited toward the total service charges. These units will be returned according to the shipping instructions provided on the Service Request form. If AP discovers that a unit cannot be economically repaired, all service work will cease and the customer will be notified.
All repairs performed on Conditional Support products will be on a "best-effort" basis subject to the availability of key parts and assemblies. In certain situations we may propose the use of refurbished assemblies from a previously salvaged unit.

AP products under conditional support

Product 	Discontinued Date 	Conditional service begins 
Sys One 	July 2002 	July 2007 
2022 	October 2001 	July 2007 
2222 	July 2002 	July 2007 
2300 	October 2001 	July 2007 
2322 	July 2002 	July 2007 
2422 	October 2001 	July 2007 
2500 	October 2001 	July 2007 
2522 	October 2001 	July 2007

http://ap.com/products/product_list.htm

Bye,

Peter

 

[bcarso]

It could become necessary to have a few Ap's around to keep one working, as I know some car enthusiasts do with beloved ancient models.

 

[clintrubber]

[quote author="bcarso"]It could become necessary to have a few Ap's around to keep one working, as I know some car enthusiasts do with beloved ancient models.[/quote]
You're right. First line of defense is to have a dual channel model, as most are, but after that you need indeed another.

 

[mikep]

[quote author="NewYorkDave"]Recently, after campaigning for it for six years, my employer bought me a Portable One +. I love it![/quote]

nice. I had one of those on my bench for a couple of years at a previous employer. lately Ive been spending alot of quality time with another kind of very spendy test gear, mainly a 30GHz HP/Agilent spectrum analyzer. too bad it only goes down to 9kHz, the resolution is pretty incredible, Id love to see how some of my audio gear looks on it.

Im thinking about buying an old Tek 7000 series mainframe for my bench and just noticed they made a spectrum analyzer plugin for it called the 7L5. anyone ever use one of these on an audio bench?

 

[bcarso]

[quote author="mediatechnology"]Sam - Thanks for the post and link. Did you see the low noise measurement preamp cite below Jung's?[/quote]

BTW, there is something a bit fishy about the Colin preamp schematic (pg. 26 of the issue). It may merely be a typo for the value of R8, said to be for d.c. stabilization of U1. The value seems too small to properly realize the desired composite first stage gain of 100 (let alone "very stable gain" as the author says). I would ask Colin but they don't provide an email addy.

As drawn the open-loop gain from J1 input to U1 output would only be about 1.5k/(1/[2gm] + 10 ohms), or about 1.5k/35 = ~42.9*. Feedback via R6 and P1 will only make the overall gain less.

R8 could be much larger and still keep the 797 happy, although some caution needs be exercised since it is a pretty high input bias current amp. If necessary a compensating R with C across could be placed in the "+" input to common.

I also note that, once there is sufficient loop gain, the input capacitance given at the preamp input is incorrect, since the feedback bootstraps out most of the input GS capacitance. There is still the substrate capacitance and the drain-gate C, but the former is probably more-or-less moving with the sources, and as Colin points out the drains see the low Z of the 797 inverting input, so only the magnitude of twice each Cdg appears, about 11pF. Note as well that the text says the input C is 33pF whereas the picture of the box shows an inscription of 72pF. Probably neither is correct, although the loop gain issue needs to be settled before it can be properly predicted.

I was happy to see from the latest datasheet that LIS, after I chatted with them (whether due to that or not), ended up bringing out the substrate connection on the SOIC version of the 389. It can be helpful to drive this in certain circuits, when one half of the part is used as a current sink or source for the other. Maybe the case on the TO-71 package version is also a substrate connection.


*Assumes the LIS datasheet spec on gm per section of the LSK389

EDIT: add brackets in expression for gain, for clarity

 

[Samuel Groner]

Did you see the low noise measurement preamp cite below Jung's?

Yes, I did. I agree with Brad that the first stage deserves some suspiciousness. I didn't quite understand why the author did not chose the rather common bias of the input FETs by DC coupling to the opamp and providing appropriate bias to the noninverting input (i.e. half of the "Cohen" mic preamp frontend). This would greatly improve PSRR and saves a few medium sized electrolytics. The choice of the AD797 seems a bit misguided as well--there is no need for low noise here (it's perfectly feasible to get a -140 dBu EIN with a TL071), the high o/l gain of the AD797 seems to enforce a wacky compensation scheme and it for sure wastes some quiescent power.

With respect to the Jung article: it is well written indeed and provides us with very usefull information. On the other hand I've got the feeling that the restriction to one measurement setup is misleading. IMO it would be important to measure "differential mode" output impedance (I just made up this terminology--I mean varying the output voltage only instead of both bias string and output voltage) as well, especially close to saturation. It might well happen that a topology performing poor in the published tests would be pretty good at "differential mode" output impedance (I think the rejected "One Vbe Current Source" would be such a candidate).

In addition to this, the text seems to imply somewhat (or at least it does not explicitly states otherwise) that the PSRR of an amplifier is mainly dependent on the current regulators used--which is wholly wrong as most common amplifier topologies have other PSRR degrading mechanisms which will completely dominate things.

Samuel

 

[JohnRoberts]

[quote author="Samuel Groner"]

Did you see the low noise measurement preamp cite below Jung's?

Yes, I did. I agree with Brad that the first stage deserves some suspiciousness. I didn't quite understand why the author did not chose the rather common bias of the input FETs by DC coupling to the opamp and providing appropriate bias to the noninverting input (i.e. half of the "Cohen" mic preamp frontend). This would greatly improve PSRR and saves a few medium sized electrolytics. The choice of the AD797 seems a bit misguided as well--there is no need for low noise here (it's perfectly feasible to get a -140 dBu EIN with a TL071), the high o/l gain of the AD797 seems to enforce a wacky compensation scheme and it for sure wastes some quiescent power.

With respect to the Jung article: it is well written indeed and provides us with very usefull information. On the other hand I've got the feeling that the restriction to one measurement setup is misleading. IMO it would be important to measure "differential mode" output impedance (I just made up this terminology--I mean varying the output voltage only instead of both bias string and output voltage) as well, especially close to saturation. It might well happen that a topology performing poor in the published tests would be pretty good at "differential mode" output impedance (I think the rejected "One Vbe Current Source" would be such a candidate).

In addition to this, the text seems to imply somewhat (or at least it does not explicitly states otherwise) that the PSRR of an amplifier is mainly dependent on the current regulators used--which is wholly wrong as most common amplifier topologies have other PSRR degrading mechanisms which will completely dominate things.

Samuel[/quote]

While I'm not defending the specific design, cap coupling FET front ends is a simple way to finesse the wide range of pinch off voltages you are likely to encounter without trims or selected parts. I have used both cap coupled FET inputs, and DC coupled in my sundry phono preamp designs over the years. FWIW my later DC coupled variant, used a DC servo loop to deal with Vg-s and DC operating point, so I guess it was indirectly cap coupled too. DC or AC coupling of the FET will impact it's LF response, but shouldn't affect CM characteristics which will be dominated by general circuit topology.

----------
Current sources come in many flavors. Different applications have different needs so I guess we all should be a little careful about generalizations that reflect our personal experience or biases.

WRT terminology, you are talking about the general characteristic of current source output compliance, but since compliance is generally just considered a voltage range for proper operation, you are drilling down into specific variances from ideal operation rather than a voltage threshold. This is not unlike parsing out specific types of distortion vs. a lumped together total number. Just like measuring distortion, how you measure will make a difference in the result.... I suspect the quirks of sundry current sources are well known to IC designers and probably well explored in those journals (IEEE).

Discussion of this topic just reminds me of how remarkable the performance of some of these new generation ICs with linearity specifications down in the -140+ dB ballpark. Discrete design is fun, but IMO difficult to meet or beat the modern integrated stuff for raw performance (at least difficult for me).

JR

 

[bcarso]

[quote author="JohnRoberts"]

I suspect the quirks of sundry current sources are well known to IC designers and probably well explored in those journals (IEEE).

JR[/quote]

Perhaps, but I don't know of a comprehensive reference. Maybe it's time someone wrote one.

 

[valvehead]

Jung's article is very informative. I had no idea that the one-vbe regulator has such poor PSRR. I'd like to see how well these circuits maintain their output current versus supply voltage and load voltage. I look forward to part two.

 

[JohnRoberts]

[quote author="bcarso"]

Perhaps, but I don't know of a comprehensive reference. Maybe it's time someone wrote one.[/quote]

This looks interesting

http://www.iee.org/Publish/Books/Cds/cs002c.cfm#7Dynamic

Not to sound like a broken record, but I'd like to pick the brain of the engineer(s) who designed those new ultra linear opamps. While they may have had some help with improved process characteristics, but you know they had to deal with this kind of stuff too, and probably much more. That kind of linearity doesn't just happen.

I suspect there's also device configurations that are hard to mimic discrete like multiple emitter structures and other oddballs that may be especially useful in mirrors. I guess the utility of specialized devices may make all discrete mirrors somewhat suspect wrt absolute performance.

IIRC at least one semi manufacturer tried to sell dedicated IC current mirrors but they didn't sell well and soon faded from sight.

JR

 

[bcarso]

[quote author="JohnRoberts"][quote author="bcarso"]

Perhaps, but I don't know of a comprehensive reference. Maybe it's time someone wrote one.[/quote]

This looks interesting

http://www.iee.org/Publish/Books/Cds/cs002c.cfm#7Dynamic


JR[/quote]

Thanks for reminding me of that book---I have it but haven't looked at it for quite a while. When I corresponded with him briefly Barrie had forgotten he'd even written his chapters!

He wrote an interesting article later on about how misused the term "current-mode" had become. I think a link is in a very old thread in here, maybe even one that I initiated. Whether it works or not now is another matter.

EDIT: Found the book (it was only several feet away) and skimmed Gilbert's chapter on bipolar current mirrors. It is excellent, as expected, but I'm espcially tickled that a possibly-new topology I happened upon recently is not there! Of course it could well be somewhere else, but it still pleases me that there was another squeeze in the toothpaste tube :grin: .

EDIT II: The book (ISBN 0863412971) can be gotten for as little as 33 bucks + S/H I see. I'd recommend it, even though it's slightly dated (1993).

 

[bcarso]

[quote author="valvehead"]Jung's article is very informative. I had no idea that the one-vbe regulator has such poor PSRR. I'd like to see how well these circuits maintain their output current versus supply voltage and load voltage. I look forward to part two.[/quote]

Although PSRR is not very good, the output impedance is quite high, in the tens of megohms typically. And how noisy are our rails, anyway?

 

[Freq Band]

Walt J. is posting over at DIYAUDIO....topic is his super regulator and other stuff. You might be able to ask him any questions....but only if it has not been covered/already answered in his articles. He expects people to read those first. :wink:

=FB=