Low-distortion Audio-range Oscillator

Jump to Latest
#8,006
jan.didden said:
Actually the naked Zfoils on offer are about twice as expensive as new ones from a trusted distributor ...

Jan
Click to expand...

Are you talking about the Texas Components VAR? Those are crap, stay away from those. Those are not the z201 you linked to earlier. Talking about

https://www.ebay.com/sch/i.html?_from=R40&_trksid=p2380057.m570.l1312.R1.TR9.TRC0.A0.H0.Xz201.TRS1&_nkw=z201+vishay&_sacat=0

A new Z201 is $30-$50.
https://www.newark.com/vishay-foil-resistors/y145310k0000v9l/res-metal-foil-10k-0-005-0-6w/dp/39T1681?st=vishay%200.2ppm

I should mention that I bought 10 new 10K VARs from Newark last year. They averaged .5 to 3ppm per/C TCR. So none met datasheet specs. Their datasheet is fiction, unfortunately.
 
Last edited:
#8,007
A fundamental question would be what does the Z foil bring to an AC circuit that the earlier generation foil resistors did not? And for that matter how would they compare to precision wirewound resistors like HP used in the HP400 voltmeters?

For DC precision and accuracy there are specific requirements but for AC they may not be the same issues.
 
#8,011
If there was a way to fine-tune the oscillator by an external voltage, a simple loopback involving ADC with crystal clock -> DSP measuring frequency -> DAC may be cheaper and more reliable than trying to make analog components stable without any control feedback. A simple/cheap 16bit ADC would do, if only frequency precision were to control.

RPi zero (5USD) + microSD card, USB 16bit soundcard for sampling (1USD), MCP4725 I2C DAC (1USD). Nonlinear curve-fitting will measure the frequency at sufficient precision. RPi frequency stability is at fraction of ppm Raspberry Pi 2 – Frequency Stability – 2N3904Blog
 
#8,016
Foil resistors are "dangerous" for AC uses because their tempco is low only at high frequencies and at DC. They achieve this low tempco by bonding the foil to a substrate with a complementary expansion coefficient, so that when the foil is heated and expands, the substrate contracts, cancelling the resistivity change in the foil.

This works perfectly at DC and also at high frequencies, but for frequencies somewhere in the middle, the rate at which the foil and substrate heat and cool and "talk to each other" becomes a factor. At some frequency, the two effects can be out of phase by 180 degrees, making the effective tempco double that of unbonded foil, which can cause 3rd harmonic distortion from signal induced thermal resistance modulation. I'm not sure what the raw tempco of their foil is, but given that we all have the same nichrome alloys to work with, it's probably not going to be much different than a traditional high quality film resistor, and may actually be larger if they chose a foil tempco higher than ±zero, to fit the expansion properties of their chosen substrate.

Typically, the frequency region where this happens is somewhere around 100Hz, so yes, these resistors won't be a problem for a 10kHz oscillator or notch, but I'd imagine they'd cease to be as marvelous as their spec sheet below 1kHz.

So, my instinct would be to use high quality films like the Susumu URG, high quality through hole films, or precision wirewounds if you can manage the stray reactance. Bruce Hofer of Audio Precision has cautioned against the Vishay foil "trick" for some time now, because of this LF distortion possibility, so I take it as something to avoid.

Jan, your idea of thermal management with a mini oven or otherwise clever packaging is also a very good idea. A tempco around a few ppm and a low temperature excursion will give both absolute stability as well as low AC distortion. Surface mount parts are easiest to use in such a system, since they mount flat and can be thermally bonded to a controlled thermal spreader very easily. Most SMD resistors you'll want to use will fit in a 0.55 mm mounting height system, making a simple plane to thermally bond to - just put only resistors and things shorter than 0.55 mm on one side of your PCB and it's easy to stabilize them. The chosen capacitors can be bought in consistent C0G SMD sizes as well, and if they can't mount on the 0.55 mm side, they can be placed together on the other side to create a thermal bonding plane of a different height. There are many ways to do the thermal bonding and control, but the basic dimensions of SMD components make it pretty easy to stabilize an entire circuit.
 
#8,018
Monte McGuire said:
Foil resistors are "dangerous" for AC uses because their tempco is low only at high frequencies and at DC. They achieve this low tempco by bonding the foil to a substrate with a complementary expansion coefficient, so that when the foil is heated and expands, the substrate contracts, cancelling the resistivity change in the foil.

This works perfectly at DC and also at high frequencies, but for frequencies somewhere in the middle, the rate at which the foil and substrate heat and cool and "talk to each other" becomes a factor. At some frequency, the two effects can be out of phase by 180 degrees, making the effective tempco double that of unbonded foil, which can cause 3rd harmonic distortion from signal induced thermal resistance modulation. I'm not sure what the raw tempco of their foil is, but given that we all have the same nichrome alloys to work with, it's probably not going to be much different than a traditional high quality film resistor, and may actually be larger if they chose a foil tempco higher than ±zero, to fit the expansion properties of their chosen substrate.

Typically, the frequency region where this happens is somewhere around 100Hz, so yes, these resistors won't be a problem for a 10kHz oscillator or notch, but I'd imagine they'd cease to be as marvelous as their spec sheet below 1kHz.

So, my instinct would be to use high quality films like the Susumu URG, high quality through hole films, or precision wirewounds if you can manage the stray reactance. Bruce Hofer of Audio Precision has cautioned against the Vishay foil "trick" for some time now, because of this LF distortion possibility, so I take it as something to avoid.

Jan, your idea of thermal management with a mini oven or otherwise clever packaging is also a very good idea. A tempco around a few ppm and a low temperature excursion will give both absolute stability as well as low AC distortion. Surface mount parts are easiest to use in such a system, since they mount flat and can be thermally bonded to a controlled thermal spreader very easily. Most SMD resistors you'll want to use will fit in a 0.55 mm mounting height system, making a simple plane to thermally bond to - just put only resistors and things shorter than 0.55 mm on one side of your PCB and it's easy to stabilize them. The chosen capacitors can be bought in consistent C0G SMD sizes as well, and if they can't mount on the 0.55 mm side, they can be placed together on the other side to create a thermal bonding plane of a different height. There are many ways to do the thermal bonding and control, but the basic dimensions of SMD components make it pretty easy to stabilize an entire circuit.
Click to expand...

Hi Monte,

Thanks for the info. Do you have any link or paper where Bruce talks about that? I know his presentation on low distortion design but I don't think he mentioned it there.

Edit: what substrate would contract when heated?

Jan
 
Last edited:
You must log in or register to reply here.
Top Bottom