Diode physics

[as_audio]

Well, Jan "was looking for a diode".

 

[MarcelvdG]

Yes, originally he did. The thread gradually shifted from diode physics to simple shunt regulator design.

 

[as_audio]

Interesting to see in #38 that Vbe is 789 mV at Ic 60 μA, but 451 mV at 200 μA,
using the same type of transistor.

 

[simon7000]

If you want a constant voltage, use a voltage regulator.

Pick a nice low power rail to rail opamp, a micro power voltage reference and an output transistor. A capacitor should clean up any noise driving the output.

 

[MarcelvdG]

A capacitor won't filter much when you drive it from a near-zero impedance and micropower voltage references are generally quite noisy, so it's probably better to put an RC noise filter in between the reference and the op-amp - but then again, Jan didn't specify noise requirements.

If noise and inductance don't matter you could use a TS4061 or similar shunt regulator IC.

 

[Elvee]

In fact it starts to look like those ic shunt regulators ...

Stop beating around the bush, bite the bullet and use a 1.2V one: it will have an inexistent dynamic resistance, zero tempco, and it's only a single part

Brain fart: how about a 3-terminal voltage regulator that could sink current?

Same as above, just adjustable

 

[jan.didden]

OK OK< stop twisting my arm ;-)

Recap: my query originated from trying to use a 2-diode drop 'thingy' as cathode resistor in a small signal tube circuit (no I won't show the circuit).

Then I looked at the diode-drop variation with current, and I thought: which type of diode would have the lowest voltage drop variation, ie. the steepest V/I slope, hence the 'physics' question.

And look where we ended up! But, as often here, I learned several things on the way, for which I thank all of you.

PS Yes there are 1.2V shunt references, but they have noise and cost upward from $6. I do have a Dutch reputation of frugality to uphold.

Jan

 

[as_audio]

Sure you can beat any tube circuit to death using a few transistors.
What would be the use of using tubes then ?
I understand why you do not want to show the circuit.
Georg Neumann invented the NiCd cell in order to replace a cathode resistor.

 

[jan.didden]

Actually you make a very good point. The tube is included in an attempt to find out what there is to this 'tube' sound.
So maybe I should put on hold my natural tendency to wrinkle out every non-linearity I can think of. In this case it would defeat the purpose.

Jan

 

[as_audio]

Traditionally you would use a big cap in parallel
to the cathode resistor, confident you know it

 

[jan.didden]

I've heard about it ;-)
The circuit has overall DC feedback so not worried about Ia drift.

Jan

 

[Mark Johnson]

... Yes there are 1.2V shunt references, but they have noise and cost upward from $6. I do have a Dutch reputation of frugality to uphold.

• AZ432BZTR-E1
  
• 1.25V
  
• ±1%
  
• 0.15 ohms dynamic impedance (guaranteed MAX spec)
  
• 20ppm/°C Typical
  
• $0.293 (qty=10)

Sales page at digikey

Only 625 of them on the shelf. Buy 25 pieces right now.

 

[jan.didden]

I think that does it. Thanks Mark!

Edit: decided to take 100. Dirt cheap!

Jan

 

[MarcelvdG]

It doesn't have any noise - well, no specified noise anyway

 

[jan.didden]

Yes that will have to be tested, although the description says 'low noise'. But it isn't specced.
The original data sheet says 'not for new designs, no alternate part'. That must be why it is so cheap.

Jan

 

[MarcelvdG]

My educated guess would be around 70 nV/sqrt(Hz), assuming that it uses a simple bandgap with 8:1 transistor and 1:1 current scaling (which is suboptimal, but only slightly) and that most of the minimum cathode current for regulation flows through the bandgap reference. But that's only a guess.