Samuel.
Nulling refers to the action you are controlling. The meter will not read zero. As you tweak the controls to get a dip in the meter reading that would imply you have nulled one of the distortions. The meter will still read the noise and any other distortion products.
It is the dip that indicates the oscillator set to the harmonic (and greatly reduced in level for the harmonic under test, all of it's own distortion and noise products) is at the same level as the oscillator under test's harmonic. It is best if the oscillator set to the harmonic frequency is as good as the oscillator under test, but it does not have to be!
ES
The overall distortion behaviour of an analyzer is probably even more complex than that of a generator, because the circuit complexity is usually higher. But the usual suspects are the same: passives, amplifiers, multipliers, control voltage ripple and layout effects.
Samuel[/QUOTE]
We know how to deal with the 'usual suspects', right? That was one of the upgrades to the HP 339A.... on both parts of it.... osc and analyzer. The results were great performance enhancement. But we were stuck with switches that are not up to the task and continue to need periodic cleaning. Future new designs should take the path Japan equipment makers have -- using hermetically sealed near perfect contact relays.
Thx-RNMarsh
Samuel[/QUOTE]
We know how to deal with the 'usual suspects', right? That was one of the upgrades to the HP 339A.... on both parts of it.... osc and analyzer. The results were great performance enhancement. But we were stuck with switches that are not up to the task and continue to need periodic cleaning. Future new designs should take the path Japan equipment makers have -- using hermetically sealed near perfect contact relays.
Thx-RNMarsh
Last edited:
Your word!
In places where the switch sits at a nice virtual earth, and voltage-dependent off-capacitance can be rendered negligible by a dual-throw arrangement with the second position grounded, it is even possible to use FET switches. Although I've not gone this route, and applied relays throughout as you suggest.
Samuel
When I look at the price of a 12 position single pole rotary switch with gold plated self wiping contacts it is $148.70 each or $12.39 per contact.
The relays I use are gold plated bifurcated crossbar double pole, double throw and cost $5.33240 ea. (I do buy these in modest numbers.)
So the preferred technology has changed enough that switches are in limited production and suitable relays are being mass produced at a significantly lower price.
ES
The relays I use are gold plated bifurcated crossbar double pole, double throw and cost $5.33240 ea. (I do buy these in modest numbers.)
So the preferred technology has changed enough that switches are in limited production and suitable relays are being mass produced at a significantly lower price.
ES
There are surely quality rotary switches at ~1/10th the price you quote, but I agree with the general message. Which relay type do you use?
Samuel
Samuel,
I use these http://pewa.panasonic.com/assets/pcsd/catalog/agq-catalog.pdf
But I have only tried about ten different types so I can't vouch for them being the best, they work well in my application which is a high reliability emergency message inserter. It uses a bank of four relays for each switch function so that a single failure will not affect the system. It gets a bit complicated as it has to switch off the main AES-3 audio stream, the special subwoofer lines and insert a signal into the analog distribution system. It also has the ability to recognize when the sound system is running on the emergency generator and reduce the system bandwidth to reduce power needs.
I also use them in my test oscillator for R & C switching.
The lower cost rotary switches I have tried were sealed silver self wiping and had issues with not all of the contacts behaving the same.
The open frame switches of course suffer from sulfur contamination. (Silver oxide is actually a better conductor than silver!)
So if you have a good rotary switch, please don't keep it a secret.
ES
OOPS I just paid a bit more attention they are silver palladium not gold! The TQ series are gold.
I use these http://pewa.panasonic.com/assets/pcsd/catalog/agq-catalog.pdf
But I have only tried about ten different types so I can't vouch for them being the best, they work well in my application which is a high reliability emergency message inserter. It uses a bank of four relays for each switch function so that a single failure will not affect the system. It gets a bit complicated as it has to switch off the main AES-3 audio stream, the special subwoofer lines and insert a signal into the analog distribution system. It also has the ability to recognize when the sound system is running on the emergency generator and reduce the system bandwidth to reduce power needs.
I also use them in my test oscillator for R & C switching.
The lower cost rotary switches I have tried were sealed silver self wiping and had issues with not all of the contacts behaving the same.
The open frame switches of course suffer from sulfur contamination. (Silver oxide is actually a better conductor than silver!)
So if you have a good rotary switch, please don't keep it a secret.
ES
OOPS I just paid a bit more attention they are silver palladium not gold! The TQ series are gold.
Last edited:
The Panasonic AGQ is special in that only the stationary contacts are gold clad--don't recall having seen this before! I'm using the somewhat less expensive, fully gold cladded, and even slightly smaller, Omron G6K. They don't explicitly mention the use of twin crossbar contacts, but the typical life cycle is about 2x that of the Panasonic AGQ and the minimum load the same, so this is probably not a problem. The Panasonic TQ is rated for another factor of 2 more cycles and half the price, but then much larger. Where a SPST contact is enough, I'm using the Meder SIL reed relay which ought to be even more reliable than above types.
The rotary switch I had in mind is the Grayhill 51 series.
Samuel
The reason why I look for crossbar contacts is that the off state normally closed contacts are used to carry signal.
The interesting data sheet information is the pull in voltage and then the release voltage. For example a 12 volt relay will actually close with just 9 volts. Now it will close slower at the lower voltage and that can be a problem when used with higher voltage and current.
For this use the more interesting parameter is the hold voltage. Typical would be 1.5 volts is enough to hold off the spring back. What I read from that is that the spring force holding the normally closed contacts under the no coil power condition is that there is only 1/8 of the contact force.
That is why the failure mode in audio use is not the mechanical fatigue given in the data sheet as life expectancy. Instead it is contact contamination of the lower closing force normally closed contacts. So sealed relays should be better than open or shielded.
In higher power relays sealed is a disadvantage as there are contaiments created by arcing and these need to get out and away from the contacts.
In a sealed relay special attention has to be paid to issues such as outgassing from the plastic case. These gases can leave a non conducting film on the contacts that is not cleaned off by the lower contact forces.
Another issue with relays is the proximity of ferrous material to the signal conductors. The closer the current carrying conductor the higher the induced distortion. It seems from practical experience that the smaller relays have less magnetic metal and less of an issue. This seems contrary to expectations that the signal conductor is close to the magnetic components due to tighter spacing.
As with all designs there are trade offs and knowing what to look for and confirm with actual measurements is a valid design approach.
So I will have to try some of the Omrons as they may be a better choice for really fussy stuff.
ES
The interesting data sheet information is the pull in voltage and then the release voltage. For example a 12 volt relay will actually close with just 9 volts. Now it will close slower at the lower voltage and that can be a problem when used with higher voltage and current.
For this use the more interesting parameter is the hold voltage. Typical would be 1.5 volts is enough to hold off the spring back. What I read from that is that the spring force holding the normally closed contacts under the no coil power condition is that there is only 1/8 of the contact force.
That is why the failure mode in audio use is not the mechanical fatigue given in the data sheet as life expectancy. Instead it is contact contamination of the lower closing force normally closed contacts. So sealed relays should be better than open or shielded.
In higher power relays sealed is a disadvantage as there are contaiments created by arcing and these need to get out and away from the contacts.
In a sealed relay special attention has to be paid to issues such as outgassing from the plastic case. These gases can leave a non conducting film on the contacts that is not cleaned off by the lower contact forces.
Another issue with relays is the proximity of ferrous material to the signal conductors. The closer the current carrying conductor the higher the induced distortion. It seems from practical experience that the smaller relays have less magnetic metal and less of an issue. This seems contrary to expectations that the signal conductor is close to the magnetic components due to tighter spacing.
As with all designs there are trade offs and knowing what to look for and confirm with actual measurements is a valid design approach.
So I will have to try some of the Omrons as they may be a better choice for really fussy stuff.
ES
Correct, a very thorough work. Also consider Slade, Paul G.
Electrical Contacts: Principles and Applications
I don't like their "no settling time"; the trouble with that is that if the gain control element (here an unspecified photoresistor) reacts fast enough, and the LED controlling it is bound to, at ultra low frequencies the gain changes in each half cycle (= distortion). I agree that the 'bounce' in level as you change frequencies with a thermistor one, like the one I've been using for the last forty years (to the point where the batteries it needed became unavailable and I was forced to build it a mains power supply), can be annoying, but not intolerable, but I'd prefer this to not trusting the distortion below a hundred Hz.
Driving the output from a pot assumes the load impedance is resistive. I'd have preferred a buffer stage (particularly since there's a spare op amp in the package unused). Distortion will be generated by mismatch in components, particularly the tracking on the stereo pot; they haven't specified close tolerances, and if the actual frequency is important (which it quite often is) you're going to have to borrow/hire a frequency counter to calibrate it.
Still, depending on your use (I certainly wouldn't choose this for looking for 0.05% distortion in an amp) it'll sound better than a function generator, and will be massively more stable and portable than the beat frequency oscillator that was the first low distortion signal source I used professionally.
Driving the output from a pot assumes the load impedance is resistive. I'd have preferred a buffer stage (particularly since there's a spare op amp in the package unused). Distortion will be generated by mismatch in components, particularly the tracking on the stereo pot; they haven't specified close tolerances, and if the actual frequency is important (which it quite often is) you're going to have to borrow/hire a frequency counter to calibrate it.
Still, depending on your use (I certainly wouldn't choose this for looking for 0.05% distortion in an amp) it'll sound better than a function generator, and will be massively more stable and portable than the beat frequency oscillator that was the first low distortion signal source I used professionally.
Yes,
If you ever look up his bio, an interesting career.
Until the US telephone network stopped being a monopoly, virtually all low level relays were made by Western Electric. They were the ones who used the bifurcated crossbar technique. The contacts were made of Western Electric alloy #2. So one's choices for those applications were quite limited. Just to limited things even more AT&T as large as it was only built parts of the switching network. NEC built parts for AT&T. So in return NEC bought parts from AT&T's Western Electric division.
With the end of the monopoly and consumer purchases of those parts significant changes have come. So it is easier to get low level relays. But we are still limited to what the manufacturers' offer. Good thing they all have folks who have read his book!
The other nice bit is the gold flash on contacts. This is very good for low level signals but not for levels above a few volts. So when you use a low level relay for even moderate voltages this blows through the gold flash to the contact material below. That is why the small signal relays are rated for currents as high as a few amps. The issue is that as long as the contact sees low power it is a good low power contact. Once used for a bit more it is no longer good for the lowest signal levels. It is sort of self adjusting to the circuit in which it is used.
ES
Last edited:
Is this a power effect or strictly to do with voltage level? Will voltages higher than a few volts with currents in the low mA range have this effect?
If you stay under 10 volts and low current (typical signal currents of analyzers and oscillators) you will not have an issue with arcing and the like. And a completely sealed relay from atmosphere is needed.
Gold plate/clad (not flashed) over silver or as I indicated is excellent. After that there are other issues in the construction that need attention; Such as the metal of the contact arm and how it is connected to the contact metal... rivet, solder, spot-welded, ferrous metals, coil fields etc) which will add distortion to the signal flowing thru the entire relay contact path.
Thx-RNMarsh
Last edited:
BTW -- Some of the Panasonic/National THD analyzers are using relays made by NEC and by Sanyu Electric Co.
Well, I recieved my service manual for the Panamax VP-7722A. And, after looking over the calibration proceedure and sleeping on it for a couple days, the old subconcience came thru for me---- recall the 2H and 3H differences between analyzers? Well, that is now likely because I used a seperate/external generator... not the internal osc which the analyzer is tuned to. Just a fraction of a percent freq difference will cause the analyzer to be off the peak and read low.
Thx-RNMarsh
Well, I recieved my service manual for the Panamax VP-7722A. And, after looking over the calibration proceedure and sleeping on it for a couple days, the old subconcience came thru for me---- recall the 2H and 3H differences between analyzers? Well, that is now likely because I used a seperate/external generator... not the internal osc which the analyzer is tuned to. Just a fraction of a percent freq difference will cause the analyzer to be off the peak and read low.
Thx-RNMarsh
Last edited:
