0
\$\begingroup\$

I've been messing around with a noise circuit utilising a zener diode. As I understand it, the white noise is (effectively) of infinite bandwidth, from 0 to a Gazoogle Hertz. It's going through a LT1011 comparator which is rated to 10 GHz.

Clearly a wire wrapped (or point to point wired) circuit board won't manage the GHz range, but what will it manage? I know they used to wire wrap mainframe computers. Those photos are total rats' nests, and must be loaded with parasitic /stray capacitance everywhere. An estimate of magnitude would be useful. Might you get a meg?

\$\endgroup\$
  • 1
    \$\begingroup\$ You can easily get quite a few megahertz with careful path management. \$\endgroup\$ – Ignacio Vazquez-Abrams Oct 12 '16 at 0:58
  • 1
    \$\begingroup\$ I've done 100 MHz logic (ECL) without problems, Had to be careful about terminations. And 20 MHz analog, too. \$\endgroup\$ – WhatRoughBeast Oct 12 '16 at 1:23
  • 1
    \$\begingroup\$ If you keep impedances low, connections very short and have a ground plane spaghetti wiring can do low GHz. \$\endgroup\$ – Spehro Pefhany Oct 12 '16 at 1:47
  • 1
    \$\begingroup\$ Commercial wire wrap boards provided a pretty good ground plane (some better than others). As long as you don't do something stupid like bundling wires together which I've seen and had to troubleshoot, 30 ga is so thin that there is very little cross-talk. High speed analog requires careful placement of component carriers and wire routing to keep connections short, but other than that it's not outrageously difficult. It's something of a lost art now, of course. \$\endgroup\$ – WhatRoughBeast Oct 12 '16 at 2:35
  • 1
    \$\begingroup\$ The stray capacitance of "spaghetti" wire wrap routing can be quite low, often better than a breadboard or stripboard. The randomness actually helps to reduce crosstalk, vs a bunch of wires or traces neatly run tight up against each other. \$\endgroup\$ – pericynthion Oct 12 '16 at 3:57
2
\$\begingroup\$

The LT1011 has 6pF of input capacitance. Zener diodes have a very high capacitance, avalanche diodes ('Zeners' of more than about 6V) much less so , but a 10V zener is still probably 20-30pF. So you've got maybe 30pF before you even start. That will cause the noise to roll off in frequency, depending on the current you are putting into the Zener.

The LT1011 will respond at maybe 2MHz if it is seeing maybe 0.5mV beyond the threshold.

enter image description here

You can measure your noise signal with a wideband oscilloscope and see what that signal from the Zener looks like- a passive 10x probe will load it a bit, but probably not off by more than 2:1.

Bottom line, I doubt construction technique makes much difference, within reason. I don't think you're going to be getting much beyond a MHz out of of the setup. If you are using a < 5V Zener diode and a moderate current, you may not even reach the high end of audio.

\$\endgroup\$
  • 1
    \$\begingroup\$ A range of 1Hz-100kHz was reported here, edn.com/design/analog/4420926/… \$\endgroup\$ – Ale..chenski Oct 12 '16 at 1:58
  • \$\begingroup\$ @AliChen Using a 12V diode. Thanks for the link. \$\endgroup\$ – Spehro Pefhany Oct 12 '16 at 2:00
  • \$\begingroup\$ BZX 85 24V Zener running in total darkness @ 30uA. Gives me a noise level of about 950mV over 1000 samples out to 48KHz. \$\endgroup\$ – Paul Uszak Oct 12 '16 at 10:15
  • \$\begingroup\$ @PaulUszak Can you pick up mains frequency in the PSD if it isn't in darkness? \$\endgroup\$ – Spehro Pefhany Oct 12 '16 at 11:01
  • \$\begingroup\$ @SpehroPefhany Not sure exactly what PSD is (something to do with probability?). No. No mains as I'm running it off a Traco Power switched supply. The Zener is glass cased and is photo sensitive, so the noise rapidly drops away with incident light. \$\endgroup\$ – Paul Uszak Oct 12 '16 at 12:02

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.