zener-phobe

Hi guys

I just started a new discrete transisitor Pre-amplifier inspired by a Mark levinson

Great until i finished regulating for 15volt rails

Got the thing slung together and I have been chasing my tail on noise ever since
only to now find it's coming from Zeners............. ooooh yes what fun

Typically my regulator circuits follow the pic -- (see pic)

Interesting is the spectral density of the noise
If i get my soldering iron close to them the density and type of noise changes - suggesting thermal fluctuations linked to electron avalanche effect as per Boltzmans constant
The problem is I have swapped Zener after Zener after Zener with no improvment

Has anyone else had a tearing of hair moment with Zeners


i love noise -- dont you ? ....ooooooooooh joy :rolleyes:
 

Attachments

  • stab15v.jpg
    stab15v.jpg
    102.8 KB · Views: 294
That design must have super poor PSRR if that regulator ain't low enough noise... Anyway, the zener should be running at > 1 mA for reasonably low noise--at low currents multi-state noise (random changes between two potentials, similar to popcorn noise) appears and also the white noise content increases. Too lazy to calculate the zener current, but intuition suggests that it might be pretty low at low output currents. I'd go for a topology where zener current is well defined and independent of output current.

Further enhancements include the use of lower voltage diodes (3x 5.1 V will give *much* lower noise) and increased filtering.

Samuel
 
All zeners are noisy, it is just that some are noisier than others. That is why they are used as noise sources. If you want low noise then bypass with a fat capacitor.

I don't understand your circuit. It seems to rely on the output transistors having large c-b leakage. Are there components missing or in the wrong place? Why have you inserted a noisy 'regulator' between big capacitors?
 
Zeners, noisy? Hmmmm Yep, they are. This is why I don't use Zeners as voltage references in regulators where noise is an issue. I prefer to use a lower conductance low noise J-fet at Idss and cascode with another of higher Vgs at Idss of the lower conductance one. Then the constant current through a constant resistor, metal film of course, gives a constant voltage. Place a few nf cap across the resistor to diminish Johnson noise. Now you have a much better voltage reference.;)
 
Hi guys

I just started a new discrete transisitor Pre-amplifier inspired by a Mark levinson

Great until i finished regulating for 15volt rails

Got the thing slung together and I have been chasing my tail on noise ever since
only to now find it's coming from Zeners............. ooooh yes what fun

Typically my regulator circuits follow the pic -- (see pic)

Interesting is the spectral density of the noise
If i get my soldering iron close to them the density and type of noise changes - suggesting thermal fluctuations linked to electron avalanche effect as per Boltzmans constant
The problem is I have swapped Zener after Zener after Zener with no improvment

Has anyone else had a tearing of hair moment with Zeners


i love noise -- dont you ? ....ooooooooooh joy :rolleyes:

I think you have a basic issue with this circuit. You need a resistor to feed current into the base of the pass transistor. Next you can use a regulation transistor to siphon off excess base current to make it regulate to the set output. I wonder if this works at all, unless you have a very leaky pass transistor? The zener noise is about 6 steps down on the issue list ;-)

jan didden
 

Salas

diyAudio Chief Moderator
Paid Member
2002-10-08 11:31 am
Athens-Greece
There are Tc compensated aerospace spec Zeners with much lower noise than standard parts by the way, but difficult to source in small number. The Jfet and resistor Norton source is a good one although better use most Idss low noise Jfets so to end up with smaller resistor possible for low impedance less prone to hum pick up if to feed low noise head amps with small PSRR. K72V would be best. The LM329 has a constant current sourced IR or red LEDS competition alternative. Preferably fed by high Vt Jfet CCS. They can better it in absolute noise, although some RC filter towards the receiver can make it a moot point.
 
Hi Guys -- thanks for your responses

That isnt "my" circuit anyhow -- it's one from I nicked from JLH also used by Armstrong Audio back in the 1970's - and for the purposes of illustration doesnt show all of it --- it's indicative only typical of zener placement

The VRs are to null out differences in the two rails to provide proper symmetry beyond the tolerances of the other devices.
I think the R2 - R5 are a compensation at start-up when the rails are imbalanced -- I think it's trying to balance the rail rise time when the unit is first switched on.
Zener current at the moment is 3.2mA which I would have thought would have sufficed.

Right -- now that's done with ----- Back to topic -- zener noise

This is interesting especially feeding power to something with as much gain and sensitivity as the twitchy pre-amplifier stages. This issue also exists within all the LM regulators - the nV root Hz is quite interesting because most regulators and zeners are not going to be forced to live at 25°C all their lives (see graph below) Most regulator packages have zeners built into their design. I've been listening to the noise coming from LM317/318s also -- that's not acceptible either -- for the purposes of quieter life in audio, i'd like to examine the power feed to drive things like MC RIAA and other sensitive stages to reduce noise even further. Is it possible to have power of battery cleanliness from a mains derived supply with todays tech?

Thanks for the links Rodeodave

Found a few zeners with supposed lower noise density http://www.semelab.com/pdf/diode/dlcc/1N4620DLCC2.pdf
but they are still frought with thermal issues.
 
Zeners, noisy? Hmmmm Yep, they are. This is why I don't use Zeners as voltage references in regulators where noise is an issue. I prefer to use a lower conductance low noise J-fet at Idss and cascode with another of higher Vgs at Idss of the lower conductance one. Then the constant current through a constant resistor, metal film of course, gives a constant voltage. Place a few nf cap across the resistor to diminish Johnson noise. Now you have a much better voltage reference.;)

This is a simple circuit I use, to start with. To get better regulation, follow this circuit with a shunt reg.:)
 

Attachments

  • reg.jpg
    reg.jpg
    79.5 KB · Views: 144
Most regulator packages have zeners built into their design.

No. Most contemporary regulators are based on bandgap references. Otherwise we wouldn't get low enough voltages for todays world.

I agree with the others that if you want a really good regulator you'll need a different design. See e.g. discrete_series_voltage_regulator_r1.pdf; this regulator achieves 700 nVrms noise in a 22 kHz BW along with superbly good regulation. It's easy to derive a complementary version for the negative rail.

Samuel
 
and for the purposes of illustration doesnt show all of it --- it's indicative only typical of zener placement
What is the point of showing an incomplete circuit which can't possibly work? We know what zeners do, and how they are used.

"I know my circuit does not work, but I don't know why. Please help me. To stop you doing that, here is another circuit which will send you off down the wrong track. Have a nice day."
 
No. Most contemporary regulators are based on bandgap references. Otherwise we wouldn't get low enough voltages for todays world.

I agree with the others that if you want a really good regulator you'll need a different design. See e.g. discrete_series_voltage_regulator_r1.pdf; this regulator achieves 700 nVrms noise in a 22 kHz BW along with superbly good regulation. It's easy to derive a complementary version for the negative rail.

Samuel

Hi - - thanks for that ---The LM series I have tried have been inducing noise - and one ends up having to make compromises -- I was looking at the circuit here at the time http://datasheetreference.com/datasheets/ti_lm317_datasheet.pdf
I just happen to note the zener embedded -- but if the 700nV is increased due to thermal -- and has more at lower registers + feeding power to something with gain -- we're back to square one.
 
Jump in the lake -- we're talking about zeners in here - not an indictative circuit - stay on topic or go away, i bet your the real pedantic type who's really proud of himself

That's not called for really.
People were trying to point out that your circuit, where you measured the excess noise, is disfunctional. The noise your measuring wasn't the zener noise. So you kind of put all of us on the wrong foot by claiming excess zener noise from a defunct circuit that wasn't zener noise at all.

jan didden
 
Last edited: