-
I found an interesting circuit (image attached), which was created by LUKASZ SLIWCZYNSKI and published on edn.com in 1999:
SETUP:
The author switched-boosts a 12VDC supply to 34VDC and smoothes it with a linear regulator. The 7805 is configured to adjust the current within 8mA to 40mA. This current aligns well with the breakdown characteristics for a zener diode with Vz of 24V. The zener diode will produce the white noise that is sent for amplification.
I've built the power conditioning part of this circuit because I cannot figure out how the author drives the zener diode into avalanche mode when Vz is 24V and the measured voltage from the 7805 linear regulator is 5.05VDC. If the measured voltage was close to 24VDC, then the zener would produce avalanche noise, which I could tune a little with the potentiometer adjusting current. The circuit uses a current rather than a voltage source for the zener diode, so maybe I am missing something here.
QUESTIONS:
1) Why did the author switched-boost the 12VDC supply to 34VDC and then reduce it to 5.05VDC with the linear regulator? I understand that the linear regulator will smooth out the ripples from the switched regulator, but the conversion of the voltages for a 24Vz zener diode doesn't seem to make sense.
2) Is there an error in this circuit's design for generating zener diode avalanche breakdown noise? If not, can someone please explain how this configuration would work? I've tried driving a 24Vz zener diode with the circuit in my lab, and it is not going into avalanche breakdown, which I expect with only 5.05VDC driving it.
Thank you for your help and time,
Mike
-
Hello Mike,
the 7805 is wired up as a constant current regulator, not as a voltage regulator. Look at the LM317 datasheet (Page 13) for a "more typical" implementation.
Kind regards, Christian -
The circuit is a bit bad in that it boosts to 34 but uses a filter capacitor rated to 35. 50V would be more appropriate there.
However concerning the 7805 part of the circuit. The 7805 regulates the voltage between pins out and gnd to 5V. In your circuit these are connected with a resistor which will draw a current of 5/R. This will then be added to the quiescent current of the regulator and sent "down" through the zener.
The three terminal regulator is not an absolute "5V referenced to circuit ground", it just regulates referenced to its own gnd pin. I hope this helps. -
Thanks for your explanations! My understanding of avalanche breakdown is that Vr is applied until about Vz@Iz for the zener diode. The avalanche breakdown commences with a surge of current. Can you help dispel this fog in my head?
How is this circuit driving the zener diode into avalanche breakdown using the current rather than Vz? All the zener diode noise generators I find online are based on voltage sources, so I am a bit lost with how things work in this design. I've implemented the circuit up to amplification. Perhaps you can suggest a testing procedure to see the breakdown by varying the current using this circuit. Perhaps, I am not testing it properly. My power supply is set to 12VDC@500mA and I have an ammeter attached to view the potentiometer changes.
Thank you again for your help.
Mike -
@ch_scr Christian, It seems that "9.3.3 Precision Current-Limiter Circuit" is similar to this design. Thanks! I am still grappling with how the current triggers the avalanche rather than reaching and exceeding Vz.
-
Think of it like this: if no current is flowing, the 7805 would see zero volts across out and gnd. This would drive the control loop to maximize the output voltage. If the zener did not conduct the output of the 7805 would reach close to 32V which is above 24V. So there is enough voltage available in the circuit for the zener to conduct, meaning that the circuit will work as intended. (A minor detail is that as the quiescent current of the chip goes through the gnd pin, you cant really talk about the control loop being active if the gnd pin current has nowhere to go)
Essentially, due to it being the threshold when the zener conducts the gnd pin of the 7805 will end up at 24V and current throjgh the zener being 5/R+Iq -
Wouldn't using a lower voltage zener, say between 7~10V and a 12VDC Power Input with a series pot/resistor to the zener to set the current and then amplify the zener noise as required work as well?
Seems like a lot of added circuitry for a noise source unless the lower voltage zeners don't produce enough noise. Also would expect the switching noise to corrupt the lower end of the noise spectrum.
Best -
see many refs on noise diodes and noise gens.
Instead of TRNG, try a PRNG Pseudo random like
https://www.sound-au.com/project182.htm
Readly made but rare:
National Semiconductor MM5837 - an 8-pin PMOS IC that was a complete digital noise source.
https://www.alldatasheet.com/datasheet-pdf/pdf/9284/NSC/MM5837.html
Jon
See -
see many refs on noise diodes and noise gens.
At a quick glance these aren't RF. The technique could be used for RF, but you'd need a pretty fast DAC with a lot of other good stuff to achieve some decent broadband RF performance.
Instead of TRNG, try a PRNG Pseudo random like
https://www.sound-au.com/project182.htm
Readly made but rare:
National Semiconductor MM5837 - an 8-pin PMOS IC that was a complete digital noise source.
https://www.alldatasheet.com/datasheet-pdf/pdf/9284/NSC/MM5837.html
Jon
See -
I understand the use of a three-terminal regulator as a current source, but the "78L05" is connected between a -34 V supply and a diode that will see -24 V when conducting.
Should that have been a 79L05 negative regulator with those voltages? -
@Kurets Thank you for the information about how the 7805 drives the zener to avalanche. I found additional information for the lm317a, which discusses three-terminal regulation, so I've learned a lot. Thanks! Armed with my new knowledge, I think I found a simple error in my implementation of my circuit, which would account for the zener not going into avalanche. I am going to make a tweak and will report the results.
-
@TimFox I measured Vout on the tl497 in my test circuit and found +32.9V relative to ground. The input voltage is +12VDC@200mA. The tl497 is configured as a step-up, positive regulator, so I don't know why the schematic shows "-34V".
-
I misread the voltage in your posted diagram as a negative due to a badly-resolved "tilde" symbol in the image.
Of course, the Zener, as drawn, requires a positive voltage.
I believe the higher-voltage diode is used since it is more of an "avalanche" device than lower-voltage units.
The usual transition between "true" Zeners and avalanche diodes is between 5 and 8 volts. -
I am still grappling with how the current triggers the avalanche rather than reaching and exceeding Vz.
You don't. There is no magic trigger, or early mode. That's why you cannot measure anything unusual.
The prose is a bit creative, claiming some special effect.
See:
https://en.wikipedia.org/wiki/Zener_effect
In general, diode junction breakdowns occurring below 5 volts are caused by the Zener effect, whereas breakdowns occurring above 5 volts are caused by the avalanche effect.[3] Breakdowns occurring at voltages close to 5 V are usually caused by some combination of the two effects.
They are all called Zener diodes for simplicity, but >>5V are actually avalanche diodes.
At very low (100uA) currents some papers talk about a break-back effect, but that is a low frequency sawtooth, and I've never seen a real diode behave that way, as the VZ would be out of spec.
You can buy 50uA zeners, so that does not seem to be a fundamental issue.
https://www.nexperia.com/products/diodes/family/50-UA-ZENER-DIODES/
A number of vendors offer MMSZ4709, 24V tested at 50uA
For an RF noise source, you want to actively avoid any low frequency effects.
When the zener is conducting, it has a quite low impedance, so I doubt it matters to the zener if it is nominally current driven, or fed from a resistor at constant voltage.
A low ripple supply is desirable in both cases.
A common circuit for Zener control is to have the zener 'provide its own voltage'. A an opamp has dual feedback paths for an output some ratio above Vz, and then Vz is fed from that output.
See circuit #1 here
https://electronics.stackexchange.com/questions/283651/opamp-zener-regulator-configuration
That has high PSRR, tho needs checking it starts reliably.
For RF noise extraction, you may also need to watch the natural RC rolloff. eg
BZX58550-C24 Zener with 5mA is 70 Ohms and 55pF , Tau of 3.85ns
PDZ24B at 5mA is 30 Ohms, max 80pF for Tau 2.4ns
ROHM UDZV24B has curves for Cz and impedance, suggests 3pF > 15V and ~ 60 ohms 1-10mA for Tau ~ 0.18ns
With very low Cz zeners, finding a current source that does not impact that would be tricky, so a simple resistor feed would have less impact on the rf noise spectrum.
-
When conducting, an avalanche diode's equivalent resistance is very low, and tends to short out the capacitance.
At voltages below breakdown, the capacitance is important and a strong function of the reverse voltage (normal bias polarity for a Zener), while the shunt resistance is extremely high.
At 0 V bias, the capacitance can be somewhat high.
(Try to keep the Zener dark if it is in a glass package: the junction area is high and photocurrent is measurable in room light.) -
I believe the higher-voltage diode is used since it is more of an "avalanche" device than lower-voltage units.
The higher voltage also helps lower the diode Ceff. Note that Ceff falls with voltage, but the impedance rises again with voltage after a broad minimum 10~15V
ROHM gives nice curves for each diode, a couple attached for their EDZ 15V and 24V models.
For RF noise a slightly lower voltage would have a higher self-roll off frequency, and give a lower power budget at a given current. Candidates would need to be tested.
-
Oh @timfox, how funny! I've looked at that voltage declaration on the schematic at least 100 times and never noticed it was a tilda!
-
Does anyone have an idea why the author of the circuit boosts 12V to 34V and then drops it to 5V? I cannot see a benefit when the 7805 is a current regulator…
-
Does anyone have an idea why the author of the circuit boosts 12V to 34V and then drops it to 5V? I cannot see a benefit when the 7805 is a current regulator…
You need more than 12V to get the 24V zener to conduct. A zener (like any diode/LED) has no inherent current limit -> One could use a resistor, or a constant current source like in this case.
The current limiter is a bit uncommon (and power-wasting) with the 7805, but not a problem in itself.
The circuit does not drop the voltage to 5V (but rather by 5V across the current limit resistor (and potentiometer)) -
Does anyone have an idea why the author of the circuit boosts 12V to 34V and then drops it to 5V? I cannot see a benefit when the 7805 is a current regulator…
Just because it has 7805 on the package, does not mean it outputs 5V.
Here it is used as a current source, because the -ve pin is not GND, it sits at ~24V and the OP pin is ~29V
The total sourced current is the sum of 7805 Iq quiescent current ( ~5mA ballpark, varies by brand and Vdrop) plus the output resistor of 8-10mA
Iq varies with temperature and voltage drop, so 7805 is not a great current source. -
Noise diodes were selected Zeners, as production parts varies widely in noise density and spectrum
Specially selected Zeners with specific bias can yield consistent noise density.
Audio, RF and microwave band diodes are available
J -
Wouldn't using a lower voltage zener, say between 7~10V and a 12VDC Power Input with a series pot/resistor to the zener to set the current and then amplify the zener noise as required work as well?
Seems like a lot of added circuitry for a noise source unless the lower voltage zeners don't produce enough noise. Also would expect the switching noise to corrupt the lower end of the noise spectrum.
Best
The higher voltage Zeners produce more RF noise, thus they get used more often.
I recall reading that the sweet spot for practical noise source use being somewhere around 20V to 28V for the Zeners.
I'd have to dig up some references to figure out what other mechanisms were involved in addition to the lower Ceff mentioned by PCB.Wiz. -
This thread peaked our interest in Zener based Noise Sources and we remembered the well known avalanche effect with reversed biased bipolar transistor EB junctions, this will become our "Noise Zener" to investigate. So we grabbed a 2N3904 and a couple resistors (50Ω {used two 100Ω in parallel} and 1KΩ), a ceramic 1000pF (for decoupling), and a Protoboard.
The setup shown uses a SA (SSA3021X+) to measure the noise power from the reversed biased EB junction. A stable good lab PS (SDP3303X) is used for bias power for the "Noise Zener".
We selected 5MHz with 1MHz BW for the SA since this was a more "Quite Zone" in out lab (we have a very noisy lab environment and did a sweep to find a lower noise frequency area).
PN1 shows the Noise Floor for the SA without the "Zener" conducting, subsequent PN2-6 show increasing Zener bias from ~10.5 to 11V.
PN7 shows the Noise properties as the SA is slowly swept (3Hz RB) and the bias supply stepped to show a stepped noise response across the SA display, and PN8 shows the Noise Noise ON/OFF across the SA sweep. Think of these displays PN7 and 8 as more of a Noise Level vs Time rather than the usual SA Frequency Sweep. Edit: Note the SA has a 10dB External Attenuator attached.
With this crude setup not sure how the noise bandwidth would behave, a better short leaded or SMD PCB setup would be required to verify the effective BW of this type "Noise Zener". Will leave for others to experiment with if interested.
Anyway, seems the reversed biased bipolar EB junction can produce a significant amount of noise when biased properly, and serve as a crude Noise Source in a pinch.
Best
-
Just for fun we hooked this up (with 50Ω terminator) to our SDS800HD and used the FFT set to 5MHz CF and 1MHz BW, average 16. The DSO noise floor in PN155 shows considerably more unwanted signals than the SA, we made no attempt to isolate those tho, and PN156 shows the "Noise Zener" fully biased (11V).
Best
-
Packaged noise sources, as used with noise-figure analyzers, comprise a selected Zener diode and a constant-current source.
The noise is switched on and off by switching the supply voltage to the source.
A 50-ohm attenuator network at the output gives a good output VSWR, while reducing the actual noise output to a calibrated value of 6 or 15 dB ENR (noise power in excess of thermal noise in 50 ohms).
Thirty years ago, when working with an RF preamplifier with an input impedance much higher than 50 ohms, we ordered a custom source with negligible output attenuation, but I don't remember how high the noise voltage was.
Before Zener sources, there was the Sylvania 5722 vacuum diode
https://frank.pocnet.net/sheets/201/5/5722.pdf
https://radiojove.net/SUG/Pubs/UFRO%205722%20Noise%20Source%20User%27s%20Manual%202020%2004%2012,%20Typinski%20(2020).pdf .
This diode was used in "saturated emission" mode, i.e. with negligible space-charge current limitation, by operating at a relatively low cathode temperature and high plate-cathode voltage so that all of the emitted electrons flowed to the plate, unimpeded by a space-charge cloud.
In typical vacuum tubes, the plate current is determined by an equilibrium space-charge cloud that reduces the E field at the cathode surface: to first order, the current in such a diode depends only on the 3/2 power of the plate voltage and is relatively independent of cathode temperature.
However, in saturated-emission mode, the current is relatively independent of voltage, but a strong function of cathode temperature.
The advantage is that in saturated emission, the shot-noise current can be calculated directly from the DC current from first principles.