Bob Cordell Interview: Negative Feedback

Jump to Latest
#261
Hi Bob,

Found it back. Not such a sophisticated test as yours, I tested the output THD @ 35W in 8 ohms, measuring THD both before the relay (amp output) and after the relay (load resistor). See below.

The differences are really small and actually counter the expectations, so there is probably some cancellation effect. But, see the very expanded scale.

Jan Didden
 

Attachments

#262
Re: Relay Distortion

Bob Cordell said:
Relay #4 was the clear winner. This was an 80-Amp automotive relay. It had a resistance of 18 milliohms and third order distortion of -91 dB on the analyzer, corresponding to 0.00002% (-135 dB).
Click to expand...

Hi Bob,

Would you be willing to share the manufacturer and part number of this relay? Was its physical configuration practical for a power amplifier application?

Thanks
 
#263
janneman said:
Hi Bob,

Found it back. Not such a sophisticated test as yours, I tested the output THD @ 35W in 8 ohms, measuring THD both before the relay (amp output) and after the relay (load resistor). See below.

The differences are really small and actually counter the expectations, so there is probably some cancellation effect. But, see the very expanded scale.

Jan Didden
Click to expand...


Thanks, Jan. This looks like its within the resolving power of the analyzer in the presence of the amplifier's distortion. Very good amplifier, BTW, with THD of about 0.002% out to 20 kHz.

Bob
 
#265
Re: Relay Distortion

Bob Cordell said:
....... I’m tempted to speculate that the specifics of the contact surface chemistry has a large role in the distortion performance, ......
Bob
Click to expand...

Very interesting experiment. Did you try a reed switch based unit in your test?

I agree the surface chemistry is likely to blame, more specifically oxide buildup with the attendant slight rectifying effect.

It should be interesting to learn whether special cleaning procedures, or factory clean, sealed contacts still do elicit this behavior.

Rodolfo
 
#266
Re: Re: Relay Distortion

ingrast said:


Very interesting experiment. Did you try a reed switch based unit in your test?

I agree the surface chemistry is likely to blame, more specifically oxide buildup with the attendant slight rectifying effect.

It should be interesting to learn whether special cleaning procedures, or factory clean, sealed contacts still do elicit this behavior.

Rodolfo
Click to expand...


I've only tried it with high-current relays so far.

Bob
 
#267
Rectifying effect? Copper oxides are rectifying, but silver oxide is not. The signal relays I usually see have silver plated contacts, and are mostly sealed, so there's even very little silver oxide buildup. A good relay doesn't degrade the signal at all. That's why you see those fancy relay-based volume controls in some high-end stuff instead of potentiometers or even stepped attenuators.
 
#268
Nixie said:
Rectifying effect? Copper oxides are rectifying, but silver oxide is not. The signal relays I usually see have silver plated contacts, and are mostly sealed, so there's even very little silver oxide buildup. A good relay doesn't degrade the signal at all. That's why you see those fancy relay-based volume controls in some high-end stuff instead of potentiometers or even stepped attenuators.
Click to expand...


Not so fast. You may quibble about the numbers, but even reputable "good" relays have the potential to degrade the signal. Don't forget, we are talking about amperes here flowing through those contacts. If you mean by not degrading the signal at all as causing less than 0.001% distortion into an 8-ohm load at a power level of only 32 watts, maybe you are in reasonable territory. These were reputable relays I measured, and if you saw the spectrum analyzer results on some of them you might not think that "a good relay doesn't degrade the signal at all."

You may think that my test is overly sensitive, but when you can find a relay that performs exceedingly well even on these sensitive tests, that is a really good thing.

I'm not a chemist, so I cannot comment on your assertion that silver oxide does not result in nonlinearity in contact resistance.

Don't get me wrong. I'm not saying that no one should ever put a relay in the signal path. You can. I'm just saying that, unfortunately, you have to pick carefully what relay you use. If you pick right, you should be fine. If you pick merely based on brand name, current rating or contact resistance, you are taking a gamble.

I can say that most of the relays I evaluated had distortion spectra that was relatively benign, namely 2nd and 3rd. But not all.

Tomorrow I'll share some results on banana connectors.

Bob
 
#269
If I remember correctly, Relays were discussed in the Blowtorch thread...
http://www.diyaudio.com/forums/showthread.php?s=&threadid=71189&highlight=
Quite a worthy bit of reading also... 😀
Addtionally, on a similar subject, GeorgeHiFi's findings that lead him to develope his "Lightspeed Attenuator" may be relavent to the relay contact "distortion" phenomina. I beleive it can be found in the first few pages of
http://www.diyaudio.com/forums/showthread.php?s=&threadid=80194
Just a few other avenues to research...
 
#270
Actually, my mistake was I did not realize the discussion was about speaker protection relays. I was thinking of line-level signal relays, or at most headphone amp relays, which see very little current. I use hermetically sealed, either normally closed or latching relays, so there's no possibility of noise being induced from the coil. Sealed latching relays with silver contacts can be found for about $1 a piece. As for silver oxide: copper oxide is a semiconductor, which is why it rectifies. Silver oxide can be used in semiconductors, but it has much higher conductivity than the copper oxide, and unless it is doped, you will find that you won't measure nonlinearity of any significance. In the case of sealed relays, the oxide layer is usually nonexistent or very small.
 
#271
I have many relays that is "defected". I put relays in amp's output. Sometimes after 2months, the amp's output becomes small (in normal listening level at start-up), if you play the amp loud, then all becomes normal again. It is the contact of the relay that have problem after 2 months or so. Changing the relay to better one can help. I notice the cap of the better relays cannot be opened (factory sealed with glue). The problematic relays usually are not glued, the plastic cap can be opened with small screwdriver.

http://www.diyaudio.com/forums/showthread.php?postid=1111027#post1111027

I tried to explain this contact phenomena before about 200 posts ago in this thread, I’ll try this time to explain what I found a bit more clearly.
First it had nothing to do with corrosion of the contacts in potentiometers and switches, it was the fact that all had a very small pin point size contact area between the surfaces. In potentiometers it is even worse because the pressure is so light as not to wear out the carbon or plastic resistance track, Switch resistors are almost as bad especially if the rotary switch is of the single leaf variety, dual leaf is better one each side of the common, but still I saw a problem even with these.
The way the problem was found, I managed to borrow for a day a very trick digital 10 gigahertz for memory storage oscilloscope from the CSIRO a massive research company here in Australia, god knows what it was worth 50k or so it made my high end Tektronix look like a toy. With this scope I was able to store and magnify the corner’s of a cd level 1 to 2 volt 1k square wave, and what I saw was an extremely high oscillation embedded in the original this oscillation disappeared once I put extra pressure on the wipers of the pots and switches with a non conductive toothpick, all I can say is that the rise time is 100’s of volts per micro second and it causes the contacts to become poorly conductive at these levels and creates a diode effect and tries to rectify the AC music signal into DC, I’ve for want of a better word called it contact bounce, you guys call it what you will but the proof was there and disappeared once more pressure was applied, in the case of RCA connectors the area is 10 to 20 times larger and the pressure much higher, so the phenomena does not seam to happen, but I think maybe it does, ever wonder why your system sounds cleaner after you’ve done a clean up of your RCA’s.
And please all you non believers, don't say anything till you've seen it for yourself, on a scope like I had the good fortune of borrowing. Not on a toy 400mhz Tektronix like mine, you won't see it.

Cheers George
Click to expand...
 
#272
lumanauw, yes those plastic caps that pop off, such relays I can understand. But the small sealed relays for low power areas should be fine. And at low current bandlimited signal risetime should not cause the bounce you describe (seems to me current risetime matters here, not voltage), so this only should be an issue with relays after the output stage of a speaker amp. Regarding the attenuator, how can one tell if it's single leaf or double leaf? I'm using Electroswitch rotary which has the mechanism and contacts sealed, and I don't want to break it open to take a look, since that would increase oxide formation.
 
#274
XLR is a better connector than RCA. Self-cleaning too, so you can use bare silver and the oxide is rubbed off (actually gold plating tends to wear off if you plug/unplug often, that's why sometimes such connectors are rhodium plated).
 
#275
Cloink,trip over the threshold again,...retarded son of a preacher man enters
Click to expand...

The Amplimo LRZ relay Jan was talking about has silver plating contacts, not gold.
It handles 5000VA, trouble is it is way too slow.
For sequential contact switching, a relay has to respond and switch much faster.
I've been using 2 multiple contact relays in parallel from around the time of Mr Cordell's EC amp paper, see quote above.
Fully sealed and gasfilled, each of 4 contacts handles 1000VA, responds in less than 10mS, manufactured by SDS in Germany.
Serial switching 2 of these relays with a timer delay places 8 contacts in parallel.
Good for 8000VA, 40 amps continuous, 160 amps peak and 800 amps/1mS short circuit. Side by side, the footprint sizes 1" x 1 2/16".
Due to the time delay, only one of two relays risks wearing down the contacts fast. Not noticed a serious degradation after 10 years of use, maybe because i'm taught to think before pressing a button.
Let's save calculated damping and shielding of relays against vibrations for the next shreak-out of insanity.

Funny thing, one of our members offered 500 of these SDS-S4 relays for €1/pc at the trade section 3 years ago. The oldest price level i have of the S4 is $10/pc in 1989, currently the ticket is at > $20/€15 each.
I seem to be the only one to have bought a stash for eternity.=> The Mall
 
#279
Banana Plug Distortion

By the same kind of testing applied to relay contacts, I have tested for distortion in banana plugs. For this simple spot test, I compared an ordinary banana plug and jack combination with an expensive one. For the ordinary pair, the jack was gold plated in its interior, but the plug was merely nickel plated. In the more expensive pair, both were of very substantial construction and fully gold plated.

Once again, 16V rms was applied through a load resistor of 8 ohms, and the banana plug combination was place in the ground line, with the spectrum analyzer monitoring the voltage across the banana plug combination. The analyzer was set so that 0 dB represented 31.6 mV (-30 dBv). Since 2 A was flowing, 0 dB on the analyzer represents a connector resistance of 16 milliohms. A -80 dB analyzer reading represents 3.16 uV. This in turn corresponds to a working distortion level of 0.2e-6 = -134 dB = 0.00002%.

The results were as follows:

The Plug 1st 2nd 3rd
Ordinary -2 -84 -87
Expensive -2 -81 -88
The Setup -10 -87 -89

Interestingly, the banana plugs showed more second harmonic than third, and the ordinary plug combination was slightly better than the expensive one in this regard. The third row of the data represents the readings produced by the setup itself, when the pigtails going to the plug combination are merely tightly twisted together. Both plug combinations show a connector resistance of about 13 milliohms. All working distortion spectra lie between -142 dB (0.000004%) and -135 dB (0.000016%). These are very encouraging numbers, but they are based on a very small sample. Both sets of connectors were brand new.

Cheers,
Bob
 
You must log in or register to reply here.
Top Bottom