Thanks for posting that link - I hadn't seen that particular page, but it reflects what is known in the upper end of audio design. It's good to see the author is still using a CRT oscilloscope.

When designing with resistors in a feedforward/feedback circuit, there are several things to take into consideration.

The greatest factor is the composition of the resistor itself. The various compositions behave differently under varying voltage, so where there is a large voltage excursion (e.g. feedback from the output of an amplifier to a virtual earth), it's necessary to specify a composition that has a low degree of non-linearity.

Another factor is package size, both from the point of view of how the potential is distributed across the resistor (related to the first point above) and also capacitance across the package. In low-cost SM resistors, a physically small size and small value resistor can be the same composition as a large size, large value part, and we are all taught to be very cautious of large value resistors for feedback circuits unless they are of a special, expensive type.

In some of my premium designs I will sometimes mix SMT for non-critical sections with through-hole resistors for the signal path and then attempt to make sure that exactly the parts that I specified are used in production. However, there can be problems with simply being brought in to design a piece of equipment and then not being in a position to be responsible for having it manufactured. For example, company bean-counters get to the product and start substituting what they think are equivalent parts that can be sourced more cheaply than your carefully specified ones. On being dragged in again to work out why the production units are not meeting the performance specs that were achieved in prototyping I usually start with the parts list and check the make and type of each of the signal-path components. I learnt to include certain power supply path components (principally bulk decoupling capacitors) in this list after one puzzling problem.

The paper confirms my own findings but I'd add that for all signal path I avoid the use of thick film resistors totally due to their high voltage coefficient. I have not seen any significantly raised distortion levels though when using either 1206, 0805 or 0603 thin film (i.e. MF) resistors. Of course any distortion arising from voltage coefficient will also vary depending on stage gain due to the higher resistor ratios and voltage distribution.

As Boswell says, you have to be VERY careful in production as I have seen far too often that thin film and thick film chip resistors get mixed together on the shop floor. Of course, purchase departments also think they are smart by saving a few bucks and then buying the wrong parts that then get mixed up in the stores bin!

I am currently designing a new distortion measuring set up (mainly a lower distortion oscillator and improved filters than my current ageing setup). Fortunately I have not only a scope (sorry Boswell, I'm solely digital now since my trusty old TEK475 finally died) but also a spectrum analyser, so it is relatively easy for me to isolate harmonics from the noise - a bit more illuminating than just using the time domain. When I've got this new set up together I could of course be coming back to post some new findings!

I have only used metal film 1% or better for about 6 years now, but I have used 0603 on a number of boards. I guess I will start using more 1206....

Boswell:

I learnt to include certain power supply path components (principally bulk decoupling capacitors) in this list after one puzzling problem.

Really? what happened?

MrEase:
TEK475 thats all I have right now....:)

Link555, post: 368584 wrote: Boswell:

Really? what happened?

Nothing very interesting in retrospect, but some years ago I spent a while searching for a distortion problem at high frequencies in an amplifier that had proudly been re-capped, only to find eventually that the type of bulk decoupling capacitors fitted in the re-capping had a peak in its ESR (equivalent series resistance) in the audio range which was responsible for the distortion. The problem was cured by re-fitting a different series of capacitor that turned out to be almost half the price of the previously chosen one. Moral: expensive is not always best.

Link555, post: 368584 wrote: MrEase:
TEK475 thats all I have right now....:)

I'd like to revive mine but cannot find a vertical output driver IC (Tek Part no. 155-0082-00). If anyone here knows where I could get hold of one of these I would be indebted!

Back to the subject of resistors, one thing that has not been mentioned is that the excess noise of thick film resistors is much higher than metal film. This could also pose problems in a critical design.

...peak in its ESR (equivalent series resistance) in the audio range which was responsible for the distortion.

Interesting so they were not varying their decoupling values?

Back to the subject of resistors, one thing that has not been mentioned is that the excess noise of thick film resistors is much higher than metal film. This could also pose problems in a critical design.

Here is a quick thin versus thick for microwave, not audio I know but relevant...
[[url=http://[/URL]="http://www.usmicrow…"]Thin film resistors versus thick film resistors how they compare application note 110 US Microwaves[/]="http://www.usmicrow…"]Thin film resistors versus thick film resistors how they compare application note 110 US Microwaves[/]

Here is one from Bourns:
http://www.bourns.com/pdfs/thnflmap.pdf

Here is an experiment getting at the 1/f noise for metal film:
http://atmae.org/jit/Articles/massiha032502.pdf

MrEase, post: 368598 wrote: I'd like to revive mine but cannot find a vertical output driver IC (Tek Part no. 155-0082-00). If anyone here knows where I could get hold of one of these I would be indebted!

[="http://www.etestparts.com/Search.aspx?Mfg=TEK&Page=15"]Tucker electronics[/]="http://www.etestpar…"]Tucker electronics[/] claim to have 1 available. I know they are located in that place over the pond, but it may be your only chance.

You presumably have the [[url=http://="http://docmesure.fr…"]technical manual[/]="http://docmesure.fr…"]technical manual[/] for the scope?

Boswell,

Thank you so much for that link - they should be showing 0 stock soon! Let's hope it arrives promptly. The scope died about 6 months ago and I searched extensively at the time and literally just missed the part (in Canada) by about 10 minutes according to the e-mail exchange. Since then I have periodically searched with no luck. What I'm not looking forward to is having to recalibrate the scope. Without a "permanent" set up it always seems to take ages to complete. Thanks also for the link to the manual - although I do have the manuals including the DM44 meter that sits on top of mine.

I am lucky, though, to also have a TDS540 (which should be useful for the recal!). Having used early generations of digital scopes I have to agree with your earlier observation about analog scopes. Old habits die hard... I must say though that the current crop of digital scopes (and even the older TDS540) are a vast improvement on the earlier digital scope efforts.

Thanks again, I am indebted and always happy to compensate with beer!

EDIT: I just checked and it's already over a year since the 475 died. I'm getting older faster than I thought! :<)

Link555, post: 368584 wrote: I have only used metal film 1% or better for about 6 years now, but I have used 0603 on a number of boards. I guess I will start using more 1206....

After getting sidetracked with my 475 problem (the part is now in the UK and should arrive on Tuesday!) I just checked back through the other posts.

As I said earlier, I have not seen significant differences between SM thin film resistors down to 0603 size. In the original paper you linked there is no data between different SM sizes in thin film but what is clear is that the 0805 Thin Film devices show no more distortion than the Thru' hole MF's. If you want to minimise distortion AND use SM then the big thing to avoid is anything that is thick film. Again, when I get time to get my new system built then this could be a useful test for me to do to check the system out. I'm currently stacked out with "paying" work so this could take some time!