Let's Talk about Current Sources

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.
Apologies to Analag for well... a hijack, but nevertheless:

I assume yoou meant the integrated Darlingtons as a replacement for the Q1-Q2 etc. pairs? Hopefully they will have similar "diode drop" than two discretes. (ime the simple diamond buffs can be picky)

Now what's been on my mind:
would connecting Q1-Q2 collectors to BASE of Q8 and vice-versa Q3-Q4 collectors to BASE of Q6 (or just reconnecting the collectors of Q1 and Q3) do any good? (like no need for base-to-base resistor, beter Vc for input trannies etc?)

..sorry Analag 8)
 
tv said:
I assume yoou meant the integrated Darlingtons as a replacement for the Q1-Q2 etc. pairs? Hopefully they will have similar "diode drop" than two discretes. (ime the simple diamond buffs can be picky)

Yes.  And they will be a better match, most likely.  In any event I think the 0.5 ohm ballasting on the big output devices will prevent things from getting too out of hand.
 
> Might help a little by ... A resistor from base to ... very low and ill-determined currents ... tempco of the current regulator diodes ... I remember the author's name ... generally good stuff.

Picky picky. This is 1965!! The 2N2222 was so new, no wonder it got mixed for '2907. Sub-ohm resistors in electronics were nearly unknown... "surely" there should be a "K" there? Typos aside, the performance was miles ahead of most 20V 5 ohm outputs on the market in 1965. And the no-trim self-biasing would not have been possible just a few years earlier... Vbe was not consistent in production.

We could also mention complete lack of protection. Hey, kept us techs employed for decades.

When the '2222 series ramped-up, engineers knew they were "better". My father kept a sample-box of beautiful gold-leg 2N2219. The 2N3713/2N3789 parts were very-very new-for-1965 also. I wonder if C. F. Andren worked for a semi factory.... ah, nope, he was at Applied Physics Laboratory, Johns Hopkins, which means he got to see and think about the New Toys before most folks.

Andren, Carl F.; Fadali, Moneim Ahmed; Gott, Vincent L.; Topaz, Stephen R., "The Skin Tunnel Transformer: A New System that Permits Both High Efficiency Transfer of Power and Telemetry of Data Through the Intact Skin," Biomedical Engineering, IEEE Transactions on , vol.BME-15, no.4, pp.278-280, Oct. 1968
Abstract: Artificial organs which are to be chronically implanted require a means of getting power into the body from an external power source with minimum risk of infection and irritation to the skin. The skin tunnel transformer is such a device whereby high level electromagnetic energy can be transmitted through the skin with very low losses and noncritical positioning. Two versions of the transformer are shown and described, and experimental results obtained with dogs are presented. Transmission over 50 watts at better than 95 percent efficiency has been accomplished with no ill effects. The use of the transformer for simultaneous transmission of telemetry data is also discussed.


"a surgical change to the normal external body geometry of the patient."
 

Attachments

  • SkinTunnel.gif
    SkinTunnel.gif
    42 KB · Views: 198
I hated to clip those long gold-plated leads off of the TO5 and TO18 cases, even though as time went on I realized shorter connections were usually better as ft's started to climb.  And there was the happy day that I made a ~100MHz FM transmitter with a 2N706---IIRC about 1964, when I was still in high school.  Not a whole lot before that I was selecting CK722s for AM broadcast band capability.

There was another cove whose designs showed up in magazines.  He evidently had access to the very newest and frightfully expensive semiconductors.  The designs weren't as good as Andren's though, although I could hardly tell at the time.
 
Hmm.. that "tunnel" thing would fit on pierced females?
Imo it would give a fresh angle to expressions like "electric lady"..


OTOH, has anybody used current mirrors as active loads instead of sources/sinks? I did that on a gaincard I started to play with last year and it sounds leaner and meaner "by ear" (didn't take any measure so far). Esp bass is more slender and "in speaker", not too bloomy.

The pdf linked @previous page (thanks btw.) has some nasty ideas to be played with.
 
Has anybody used current mirrors as active loads instead of sources/sinks?

A (standard) CM is very close to the two-diode CCS (Q1 in the first post). Replace one diode with a resistor equal to R2, and you're done... Output impedance is the same, but PSRR got worse.

Samuel
 
ism-fig6a.gif


Is there a generally accepted way of temperature compensating the one on the right? The obvious method would be to use an NTC thermistor for R3, but that would only work over a relatively narrow temperature range (which might be enough, depending on the application).

JDB.
[Come to think of it, would an NTC have a lot of excess noise compared to a metal film resistor?]
 
Is there a generally accepted way of temperature compensating the one on the right?

Temperature compensation may lead to the concept of the Wilson mirror, and that's what I've seen in ICs (using current ratios higher than 1). PSRR suffers, as mentioned above.

Samuel
 
The "closeness" doesn't explain the audible difference I was getting.. repeatable in diff circuits.

The "closest inbetween", made by replacing the D1 with a resistor (same value as R2**), I have seen in many circuits. Quite probably even much worse psrr.


**NOT used as a mirror, but a source/sink.
 
NOT used as a mirror, but a source/sink.

It's just a matter of definition. Both CCSs from the first post will work as mirror--e.g. Q1 mirrors the current set by R1. Not with a ratio of unity, not very accurate and pretty nonlinear (that's what improves PSRR), but nonetheless a working mirror.

The "closeness" doesn't explain the audible difference I was getting.

Of course you'd need to give implementation details if you expect further insight. There are subtile differences regarding noise, transient and thermal performance of the circuits if they are worked towards a linear unity gain mirror. But impossible to tell without schematic (incl. surrounding circuit).

Samuel
 
Well - I'm not ready to publish the circuit(s), BUT there is an easy way to test: a simple BJT emitter follower buffer, loaded with an active sink (of different types mentioned, all targeted .. say @2mA).
 
just came across this: (LT3092)
http://www.linear.com/pc/productDetail.jsp?navId=H0,C1,C1003,C1040,C1055,P88400
I can't wait to get me hands on these, and they might even be useful for audio.

mike p
 

Latest posts

Back
Top