Split & Single Rail PSU Question

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.
Teekay from TK audio has a good three rail design using 317, 337, and 783 (or lm7506). It was used for various Neve projects. Another thing to consider is the typical dual rail and 48v phantom supplies out there that can easily be converted so that the 'phantom' supply rail can be whatever you want. I've done this for LED's and digital power lines.
 
apzx said:
..... Essentially, what I'm thinking here is to use an appropriately rated bridge rectifier and followed by the filter capacitors.....
You forgot a rectifier in your schematic.
 
> a third, single ended power rail

Where does it want to be grounded?

These things often work in isolation but smoke when you run audio interconnects.
 
That doesn't look like an oscillator, except by finding transistors with a rare imperfection.

I sure don't see why it can't be done with single 15V. Or really why it needs many mA.
 
apzx said:
The transistor is used in a reverse avalanche condition. Kerry D. Wong did a short little article on it. The oscillation time is set by the the RC time constant set by (in my simple schematic) R1 & R2 and C1. This makes a sawtooth-ish oscillator. I've seen this particular configuration be called a "Negistor" as well, at least it seems to pull more results on the Googles.

And yeah not every single NPN transistor will actually work, but in my quick test board I made I just grabbed a random 2N3904 I had lying around, 10k pot, 1k resistor, 10uF elctro, and a white led. It worked without a hitch, but I did cut the transistor's base lead off (apparently that can help in making the transistor work in this configuration more readily). I had to supply it with about a minimum of ~+12.5VDC for it to start oscillating. Any lower and the circuit would not begin and as I increased the supply voltage it started going faster and lowering the voltage made it a bit slower even to the point where it would not oscillate at all. I believe the main reason for the current being so high is because with say a supply voltage of about 12V and a 1k resistor gives ~10mA or so flowing through the led. Then adding some support circuitry to the mix and the current goes up a bit.
I have never heard of using that, and I like discrete transistor tricks. I found the Wong references but that's about it. It seems too process dependent to rely upon for production unless a manufacturer sells selected parts for that attribute.

JR
 
apzx said:
Without a doubt the parts would have to be selected because it relies on the B-E breakdown which is probably not very controlled in the manufacture of the actual devices. Though enough transistors seem to work just fine in the actual circuit. In trying to understand some more about how these things actually work (just for curiosity sake more than anything) I found a paper from Raymond Arthur Kjar at Iowa State University titled Avalanche mode operation of transistors from 1964. In my searches the most common thing I've found about this particular configuration is in making a circuit with extremely fast fall time and making something with an extremely fast rise time is also possible. It is an interesting circuit if all sorts of flawed.

forward avalanche is well known (I think) and a typical consideration in power amp design... (secondary breakdown, etc).

Reverse avalanche is far less common, while some bipolar transistors are designed to be used backwards for lower saturation voltage when used in audio mutes, that application is nowadays done with JFETs.

JR

PS: A diac is another semiconductor device targeted to take advantage of it's breakdown threshold.
 
apzx said:
Indeed, anything I've read on power amp design forward avalanche is taken into consideration or should be at least.

Now, before I saw this I was not aware of a BJT being able to be used in this condition. Frankly, just looking at the schematic and I I would just think that the capacitor would charge up and cause the led to illuminate briefly and that is about it. But having built it and trying to get a better understanding of things it seems to behave more like a relaxation oscillator.

Circuits like this remind me of the sheer ingenuity of some folks. A couple more that I've scratching my brain with are VCFs in particular the Steiner Parker filter wherein he used 8 diodes and their dynamic resistance with respect to their forward voltage. The other is the ARP 4072 filter and its uses a LM3900 (I believe in another thread you mentioned something about Norton Opamps) plus a bunch of differential pairs to operate. Still trying to wrap my head around how that one works.
The Norton amp is pretty much the back half of an op amp with the input long tail pair bypassed. They do somethings well mostly due to the shorter path (with less delay) but fell out of favor when real op amps are so cheap and widely available.

JR 
 
I used one in the 60's; I made a foot-operated bass organ. Sounded awful (lots of clicks).
A little later, as I was an intern, my supervisor asked me to investigate the possibility to make a clock for an arctic event recorder. He fancied that an UJT would better behave than a crystal at -40°C. Of course it just didn't...
And I never crossed path with any since...
 
Back
Top