advertisement


LM338K equivalent?

+1 and don't worry if they are small values of 1v apart; it's to be expected, will have utterly-zero effect in use.



Sidenote on having a constant-current test supply in the bits-box:


One of the more useful things I did out of bits-box junk a few years ago was to make a constant-current test thingummy out of a spare 30v wallwart. Add a simple diy constant-current source (say the simple, very reliable ring-of -two-transistors kind) at say 10mA. You can add a switch or a second such CCS for 50mA or more if you like*

As an output, I have that linked to a spare 8-pin DIL socket, with each side set of pins paralleled, so I can push-in found LEDs, Zeners, diodes, reverse-biased transistors, or whatever ... and use a dvm on the parallel pins to measure the voltage. Also to check bulbs, fuses and so on.

* Make sure the pass transistor can survive full dissipation of supply V x set current I ;)


For checking low & sub-ohm resistors it's a solid way to go - but instead use a beefy 5v wallwart, an LM317 set up as a CCS for 100mA (or more, with heatsink) - and captive croc clips on short leads, from it to the resistor lead ends. Measure the voltage across the body as close-as, inside those leads, and divide by current (0.1, for this example) - hey presto, actual resistance.

You'll need to fiddle with the 317 setting resistor to get fine current accuracy, set prior - and not want it to drift with heat - so a set current of 0.1A is good enough for many purposes, and easier to finesse (it'll be v close to 12.5 ohms; make it up with say a 15ohm 1w resistor, in parallel with 100ohm/1w, and add a 1/2w, 500ohm multi-turn preset in parallel.)

NB 0.1A and say 0.05ohms will (should!) read 5mV, for example - subject to how good contact is, and the limits of you DVM (always read the manual). If you need wider range, or closer accuracy, you'll also easily work-out how to do that - and the potential issues with it, such as current-set accuracy under full vs no-load, dissipation in the whole CCS rig, and similar. HTH.
 
Had another play with this PSU yesterday, whilst I wait for the Zeners to arrive...

I tried again to better tune the snubbers using the 'cheapo-modo', and the 50Hz spike response changed. I can't seem to get rid of it, but it did reduce slightly. This is quite annoying, as it does look like either the Mosfets/ rectifier or the traffos are ringing.

I've got some different Mosfets on order, perhaps that will make a difference, although it could still be some dodgy switch mode PSU somewhere
 
Last edited:
Pack of 10 Zeners arrived. I found 1 that was 10.8V and a pair that were 10.96. All the other were either slightly over 11V or on value. Fitted them in place of the resistors and I now have +/- 12.25V.

The +ve rail is solid, but the negative still wanders. Every so often it drops down and then creeps back to 12.26 or so.

Perhaps I should try adding a diode for temperature compensation? First I'll try a 75R in place of the 100R, as I have a bag of them.
 
I must really be loosing it.. The 75R sounds more natural, and this is replacing a 100R Kiwame with a NOS Russian item. Nothing that I'm measuring is any better/ worse. Maybe the DC offset is more stable then before...maybe
 
one low voltage zener will have a lower dynamic impedance whn run at c 10mA. The cap in parallel only helps : )
As a further development on this PSU I've recently replaced each of the Zeners with 5 large green LEDs, giving exactly 10.7V and 12V at the rails. The Zeners will not be going back.

There was a hardness to the amp that was annoying me and now it is gone, greater details / depth of soundstage are apparent.

What might have happened here? Why isn't the bypass cap able to ground all the noise from the Zener? I would have thought that 5 LEDs in series would present a higher impedance than a single zener diode, sonically/ subjectively the amp is much improved running like this.

Next test will be 3 white or blue LEDs
 
Last edited:
Don't bother, green are lowest noise. Assuming they're not just in a green lens.
Thanks, good to know.

Did I read somewhere that if they are potted/ blacked out then noise is further reduced?

These are chunky old Soviet items, I got sent 10 as a gift with the NOS 0R6 power resistors
 
Depends on several factors. If running the string of LEDs at 10mA, they could be about 5 ohms each so 25-30ohms for the string. The Zener will likely be higher at circa 70-100 ohms, but again depends on the current throughput.
 
When Joachim and Co were designing the paradise they tested a whole range of options for the ccs in the shunt, of all the topologies tried the leds had the lowest noise and the quietest ones were the 2mA kingbright pale green. I can look up the part number if you want, I'll have it on file
 
When Joachim and Co were designing the paradise they tested a whole range of options for the ccs in the shunt, of all the topologies tried the leds had the lowest noise and the quietest ones were the 2mA kingbright pale green. I can look up the part number if you want, I'll have it on file
Thanks Simon, yes please. Might be the case that what I have now is just as good. Only one way to find out.

Was there any research done on shielding them from ambient light?

For neatness I could maybe use 5 sections of a 7 segment display, will likely have to watch the current draw. Looks like I can't - common earth
 
Not that I know of, but I've never read anything that would suggest leds exposed to ambient light produce more noise that leds exposed to themselves.

Much like our eyes sense photons, but I can't shoot frickin lasers ;)
 
LEDs are photodiodes, so will respond slightly to changes in ambient light. If the light is flickering, like say a fluorescent, this will appear as noise. The photocurrents will be tiny, and if they are biased with a mA or so, will make a negligble difference. Remember, we have chosen an LED because it has low dynamic resistance - the voltage changes very weakly with current. Once you have a lid on, the board should be in the dark anyway!

I can have a go at working out how big the photo-current might be, but it is going to be really small.

To convince yourself, try connecting the LED to a DVM on the most sensitive current scale, and exposing it to ambient light, then blocking the light. I would be amazed if it exceeds a microamp.
 


advertisement


Back
Top