| Author |
Topic: Clamping Diodes |
Keith Hilton
From:
248 Laurel Road Ozark, Missouri 65721
|
Posted 17 Feb 2000 7:45 am |
|
In another post I brought up the protection offered by clamping diodes on the + and - inputs, and the outputs. These protect against high voltage spikes destroying the junctions of transistors. Again--I'm not talking about surge protectors you can buy, like computer surge protectors. So don't bring up the surge protectors you have bought, because we are not talking about power supplies. I am talking about thousands of volts generated by walking across a carpet.
I could not get my clamping diodes to work without noise. Well guys, I was totally WRONG! Hartley Peavey was RIGHT! So was Walter Jung, author of IC OP-Amp Cookbook! Use silacon switching diodes 1N4148, which is the same as 1N914, or Radio Shack 276-1122. When I clamped to the power supplies I got no noise. In the Jung IC Op-Amp Cookbook, he describes a method of clamping between the + and -, instead of to the power supplies. Maybe that was where I got screwed up. Sorry!
------------------
|
|
|
|
Keith Hilton
From:
248 Laurel Road Ozark, Missouri 65721
|
Posted 17 Feb 2000 6:52 pm
|
|
The book I have been quoting out of is really the bible used by the people who manufacture OP Amps--chips. It is a SAMS publication. Title: IC Op-Amp Cookbook 3rd edition. Written by Walter G. Jung. Hartley Peavey turned me on to this book. I do not agree with Mr. Jung on page 149 ,concerning preferred over voltage protection of the audio inputs for unprotected OP Amp inputs. He shows two signal diodes run in reverse, in parallel,between the plus and minus inputs. This causes a lot of noise! The breakdown on a diode is around .7 volts. He also shows Zener Diodes in series between the plus and minus audio inputs. This causes even more noise in the audio signal. On page 152, Mr. Jung uses the method Hartley Peavey uses. That is, where the diodes on the input, and output are clamped to the power supplies. There is no noise with the 1N4148 diodes using this method.
This is really important stuff. I would bet that about 80% of failure of effect units, pre-amps and amps come from junction failue, from over voltage, or over current on audio inputs or outputs. Seldom do you see failure in the power system, because they usually are regulated with voltage regulators. Steel players have a lot of stuff in line and many use two amps, or are tied to the PA. With the tip of a 1/4 jack it is easy to make a direct short. Also the use of the many things in line, creates supply sequencing problems ,since each device has it's own power supply. They can be turned off and on in many different orders. This can suck power in or out of the rest of the stuff. With several different power supplies, you wind up with DC off-set and differences in chassis potentials. Besides this, thousands of volts of static electricity can be built up by just walking across a carpet.
You may think I am going overboard, but I see so many failures because to many manufacturers are not protecting their stuff correctly. Hartley Peavey's stuff lasts so long because he has everything protected really good.
------------------
[This message was edited by Keith Hilton on 17 February 2000 at 06:54 PM.] [This message was edited by Keith Hilton on 17 February 2000 at 06:57 PM.] |
|
|
|
Jim Cohen
From:
Philadelphia, PA
|
Posted 17 Feb 2000 7:15 pm
|
|
"Clamping Diodes"? That sounds like the name of an alt-rock band to me!  |
|
|
|
Richard Sinkler
From:
aka: Rusty Strings -- Missoula, Montana
|
Posted 17 Feb 2000 7:31 pm
|
|
Man, I hate it when my diodes get clamped. OUCH!!!!
------------------
Carter D10 8p/10k
www.sinkler.com
|
|
|
|
Rich Paton
From:
Santa Maria, CA.,
|
Posted 18 Feb 2000 1:13 am
|
|
Keith, I fully agree with you on the use of the zeners, as a zener is inherently noisy, to the point of one being suitable itself as a noise source in testing applications. (Very common in high frequency RF and microwave systems). But I don't see where the parallel reverse limiter where its grounding is properly addressed provided the normal signal voltage isn't high enough to foreward bias the diodes.
Any D.C. leakage through the signal path, in an audio device using F.E.T. switches is a failure that simply hasn't occured yet.
It's often associated by using electrolytic coupling caps with insufficient working voltage, a common economizing measure. This isn't seen on a Peavey or other quality equipment for the obvious reason!
Your reference to clamping a signal to the power supply rails reminded me of an adage often heard in my military tech school, "B+ is signal AC ground". |
|
|
|
Mark Amundson
From:
Cambridge, MN USA
|
Posted 18 Feb 2000 6:30 am
|
|
I agree with keith on having a copy of Jung's op-amp bible (I have a mid-70's vintage version), and even his audio-specific version too. Also, the switching diodes (1N914, 1N4148, etc.) have to be reverse biased to the split supply rails (ala Hartley's application) to be of any use. The back to back diode clamping across the op-amp inputs seems hardly useful for spike protection.
There is one drawback to these diodes between the rails and the input signal trace. Any power supply trash (high frequency noise, RFI) on the supply rails will be injected onto the input. Those diodes in reverse bias act like small, picofarad sized capacitors across your input. With a good, stable, regulated power supply, they will help your inputs be free of noise, but with unregulated or battery powered designs the diodes could do more harm than good. It all depends the rail filtering.
Another point to make is that most all ICs (bipolar or CMOS) have some form of ESD protection in them, in the form of clamping transistors if the signal pins go beyond the Vcc(Vdd) or Vee(Vss) rails. Different devices have differing levels of protection with generally bipolar devices having the best self-protection to ESD.
|
|
|
|
Keith Hilton
From:
248 Laurel Road Ozark, Missouri 65721
|
Posted 18 Feb 2000 9:40 am
|
|
Rich, I wish you could see the diode protection diagrams in Jung's book. I probably have not made the hookup clear. If you like, I will make a copy and mail you the diagrams.
Mark, I totally agree with everything you said. You can get rid of the RFI you talk about, with a 39pf capacitor to ground, connected to the input. This is the way Peavey does it. I agree that these diode clamps to the power supplies do little good, unless they go back to a "REGULATED" power supply. If they didn't, the excessive voltage could just over voltage the power inputs on the chip. I agree that many Op amps have lots of built in protection. The trouble is, every OP amp does not cover every possible falure mode. When you deal with the public, and make electronics, you "HAVE" to deal with every possible failure mode. Murphy's LAW!
------------------
|
|
|
|
Rich Paton
From:
Santa Maria, CA.,
|
Posted 19 Feb 2000 10:34 pm
|
|
Keith, I went back & re-read through some articles Jung wrote for Audio Amateur Magazine in the 80's.
I apologize for commenting without a full grasp of the subject, please keep the info coming.
You mentioned dealing with the public and making electronics. Do you manufacture a product? What?
|
|
|
|
Keith Hilton
From:
248 Laurel Road Ozark, Missouri 65721
|
Posted 19 Feb 2000 10:56 pm
|
|
Rich I make the Digital Sustain Units, and the New Infrared Foot pedals that don't have pots. My infrared pedals are not the same technology as the old light beam pedals.
After answering your question, let me get back to clamping diodes. When you plug into a device ,you have a plus input and a ground. At the other end you have a plus output and a ground. I'm talking about guitar cords. To prevent electric static discharge from japing transistor junctions, seems to me you would need 6 1N4148 diodes. 2 on the plus input, 2 on the ground, and 2 on the output. These would go to the plus power supply and minus power supply. Isn't that the way everyone else sees it?
Mark e-mailed me a letter and talked about a reverse biased diode from input to output on the voltage regulator for back bleed. This really confused me? Help?
Also, Mark said the clamping diodes should be run back to the power supply "AFTER" the voltage regulators. Well, that seems strange! Looks like if you were trying to get rid of high voltage from ESD, you would want it sent to the lines before they went to the regulators. I can't quite picture why Mark said to put them after the regulators? Help?
------------------
|
|
|
|
Mark Amundson
From:
Cambridge, MN USA
|
Posted 21 Feb 2000 7:56 am
|
|
Help is on its way via e-mail, Keith.
|
|
|
|
Blake Hawkins
From:
Florida
|
Posted 21 Feb 2000 10:17 am
|
|
Keith, the backwards diode across the voltage regulator is to protect the regulator from the charge on the input filter caps at power down.
It safely discharges the caps.
|
|
|
|
Keith Hilton
From:
248 Laurel Road Ozark, Missouri 65721
|
Posted 21 Feb 2000 4:04 pm
|
|
Blake, are you saying put a 1N4001 diode across pins of a 7812 positive voltage regulator, and 7912 negative voltage regulator? What is confusing me is which pin to which, since each has 3 pins. Is it from ground to line in or what? Help?
------------------
|
|
|
|
Terry Downs
From:
Wylie, TX US
|
Posted 21 Feb 2000 9:01 pm
|
|
Keith,
The only reason why you need protection diodes from input-to-output of a series linear regulator is when the regulator OUTPUT (load) energy storage exceeds the energy storage of the INPUT. If the power is removed and the input capacitors discharge before the output capacitors do, the linear pass transistor in the regulator becomes reverse biased. If the difference exceeds the reverse breakdown of the device, it will surely fail. Most series linear regulators for audio applications have more energy storage on the input side and do not require the diode. However, some very low ripple designs employ large capacitors across the reference pin like in the case of the LM111 positive series linear regulator. It too has a reverse potential limit. Addition of capacitance at the reference terminal increases ripple rejection by 20-30dB.
Also the reason you use clamping diodes to the output of the regulator is to keep the inputs of an op-amo from going any more than a diode drop away from the op-amp supply pin.
In regard to the rationale of zener vs. 1N4148 devices is the difference between semiconductor physics of a avalanche device vs. a depletion device. The small jucntion clamp approach is widely used in many digital ESD protection schemes as well. It is a staple of the industry.
I hope this helps.
Regards,
------------------
Terry Downs
http://nightshift.net
terry@nightshift.net
|
|
|
|
Keith Hilton
From:
248 Laurel Road Ozark, Missouri 65721
|
Posted 21 Feb 2000 9:57 pm
|
|
Thanks Terry! I see what you mean about running the 1N4148 diodes back to the output of the voltage regulators. It just seemed to me running back to the7912 and 7812 voltage regulators, you should go to the line in. Because if they took a high voltage electric static discharge, it would go into the regulators. If you run the 1N4148's to the outputs of the regulators, either the plus or minus input on the OP Amp will take the hit. Here is something interesting to think about, concerning where the clamping diodes should be run to. On my foot pedal, after the voltage regulators, I go to a DC to DC converter. Sometime I will explain why I use a DC to DC converter. Anyway, when I try and run the clamping diodes back to the output of the voltage regulators, I get lots of noise. So, it appears that the clamping 1N4148 diodes will have to be run to where the power comes off the DC to DC converter. I suppose it has something to do with mixing the output of the voltage regulator, which is 12 volts, with the output of the DC to DC converter, which is 15 volts.
Mark brought up a interesting thought. He said Peavey is one of the few that even uses these saftey clamping diodes. The other guys figure the input and output caps will stop DC hits, and input and output resistors will lower the surge current. These clamping diodes by the thousand cost only 2.5 cents. So for .10 cents you could have a world of protection.
My problem has been getting them run to the right location without noise. On one of my units, power comes from not only a wall wart, but battery power is also available on the same unit. That became a routing problem, because I had a feedback on battery power. I had to run my battery power in a different location, with a 1N4001 diode.
------------------
|
|
|
|
Terry Downs
From:
Wylie, TX US
|
Posted 23 Feb 2000 7:12 pm
|
|
Keith,
I think the reason why you get noise with the clamping diodes conneted to the switching regulator is because the switching regulator output has the the switching frequency on it. A reverse biased 1N4148 is a capacitor that could couple switching regulator noise right into the signal. We get spoiled from not needing really clean supplies today with op-amps having a 90-100dB of power supply rejection ratio. Coupling power supply noise into the input signal with a capcitor is virtually what you are doing with a 1N4148.
As far as the regulators "taking a hit" of ESD, the ESD is high voltage but not enough energy to change the supply filter voltage enough to damage anything. The supply is low impedance at ESD spectra because of all the capacitance. The regulator is only low impedance with a sink at DC. You can raise the output with DC and see a very high impedance into the regulator.
------------------
Terry Downs
http://nightshift.net
terry@nightshift.net
|
|
|
|
Keith Hilton
From:
248 Laurel Road Ozark, Missouri 65721
|
Posted 23 Feb 2000 8:55 pm
|
|
I connected the 1N4148 diodes to the same point where I pickup plus power, and minus power for the chip. There was no noise. You see, I build my own power supplies. There is several places I could run the 1N4148 diodes, when I went back to the power source. What really confused the issure was on one device I had battery power or AC wall wart power. On another device I had the DC to DC converter. I have it figured out now. The secret is to not go to far back in the power chain.
------------------
|
|
|
|
