Using 1% resistors in an amp as apposed to various tolerance resistors

[Luxman1050]

I'm just wondering if I were to replace all resistors in a 60 year old amplifier with 1% tolerance resistors as opposed to 10% or 5% would this affect the voltages and current in the amp in relation to the leeway given by 10% or 5% tolerances. I'm thinking there must be a reason why 1% is never really used. Cost comes to mind.
Just a thought.

[Richardgr]

Manufacturing processes, that is all.

There are not so many places In a tube amp where 1% tolerance made sufficient difference to merit the extra cost

[Craig Sawyers]

Even 35 years ago, in high ticket Fluke gear the BOM specified metal film 1% 100ppm in critical locations and carbon film 5% elsewhere.

Now metal film resistors are cheap, and the temperature coefficient is 50ppm.

60 years ago metal film resistors were expensive and used in very expensive test gear or aerospace, and carbon film or carbon composition for genera purpose use.

There are few downsides to using 1% metal film resistors in your amplifier.

But you should only do that if the old resistors have drifted high in value.

Craig

[Luxman1050]

So using 1% metal film would cause what problems?

[McMurdo]

5 or 10% does not necessarily mean the original resistors are always going to be 5 or 10% out, its a worst possible scenario. Many of them may already be within 1%. Your query is like asking if a properly fitted suit is worse than one that nearly fits you and you wear it anyway

[broadgage]

Originally Posted by Luxman1050

So using 1% metal film would cause what problems?

None, in my view, apart from extra cost which is now trivial.

[Sideband]

There is no point other than perhaps a psychological one in using 1% resistors in a valve amp. Valves are not so critical as transistors and will work sufficiently well with 10 or even 15% voltage variations. There might be some advantage in perhaps a high-gain pre-amplifier stage where carbon resistors might increase noise levels but that is nothing to do with resistor tolerance, only the construction.

It's up to you of course but if you are expecting a huge improvement in performance, it won't happen...the valves will probably be noisier than the resistors!

[GrimJosef]

Originally Posted by Luxman1050

So using 1% metal film would cause what problems?

I suppose you might be unlucky and go from a situation where loose-tolerance components just happened to hit a 'sweet spot' in terms of performance which tighter-tolerance ones couldn't reach. But in almost all circumstances, while the difference might be measurable, it would be very unlikely to be audible.

By way of an example, consider the phase-splitter in a Leak TL/12 Plus amp. The circuit can be found here https://44bx.com/leak/Leak/Circuits/TL12Plus3.gif. The phase-splitter is made up of the two triodes in V2, an ECC81. The triode connected to valve pins 1, 2 and 3 has its grid driven from the previous stage. The other triode, connected to pins 6, 7 and 8, has its grid AC-grounded and its cathode driven from the cathode of the first triode. This asymmetry means that the audio signal generates slightly different current variatons in the two triodes. But what we really want is for the phase-splitter's output voltages to be near-perfectly matched (assuming that the output valves are also perfectly matched). The slightly different phase-splitter currents can be compensated for by giving the two triodes slightly different anode loads and we can see that Leak do this. The second triode has a 100k load resistor connected to pin 6, but the first triode has a 91k load connected to pin 1. All other things being equal, these values give more closely matched output voltages from the splitter than two 100k's would.

But if all other things are not equal - say one of the ECC81's triodes has a slightly higher transconductance than the other - then the load resistor values for a balanced output will not be 91k and 100K. They might instead be 93k and 100k. If you use 5% tolerance resistors then you might be lucky and find that their values are indeed 93k and 100k which will lead to your amp working particularly well. If you now replacee the 5% resistors with 1% ones then the 91k will never be able actually to be 93k, so in this case you'll worsen the performance.

I have heard of perfectionists sitting down with a handful of 5% resistors and fitting them one after another until they find one which causes the amp's performance to measure as well as it possibly can. I'd take a lot of convincing that it's worth the effort though.

Cheers,

GJ

[Radio Wrangler]

Wider tolerance resistors DO NOT give you more leeway (unless you are the resistor manufacturer!)

Quite the opposite. Wide tolerance resistors eat up some of the leeway that your circuit design has. You have a higher risk of problems with wide tolerance resistors than with tight tolerance ones.

When the cost of 1% resistors had fallen to the point where the extra cost was slight, HP decided that they might just as well use 1% as their standard parts. If they kept using 5% and 10% parts, then that would have meant more stores bins and more floor area, more office work in controlling stocks, testing incoming parts, quality control etc. Not only that the risk of confusion. 1% parts worked out cheaper, considering that we already needed them as a precision test equipment manufacturer.

Not mentioned so far is that the natural series for 1% parts is "E96" meaning 96 different values per decade where the upper tolerance limit of one part butts up precisely to the lower tolerance limit of the next value. So the resistors are on a series where the ratio between adjacent values is the 96th root of 10.

That's a hell of a lot of resistor values.

So we picked the nearest E96 values to the usual E24 series of 5% parts.

Also, you'll find 1% parts usually have lower temperature coefficient.

David

[trobbins]

You'd also run the risk of damaging PCB pads and traces, or tag boards, or heat stressing nearby parts, or fitting incorrect resistor values.

I'd suggest there are modern ways to test and then tweak an amp's performance, and spending time and effort there would be better for you. That test and tweak loop may involve changing resistor values, but it would be based on a valid reason.

[Radio Wrangler]

Also the materials used to make higher stability resistors are also the ones which have lower noise.

David

[murphyv310]

Originally Posted by GrimJosef

I have heard of perfectionists sitting down with a handful of 5% resistors and fitting them one after another until they find one which causes the amp's performance to measure as well as it possibly can. I'd take a lot of convincing that it's worth the effort though.

Cheers,

GJ

Hi.
Sledge hammer and walnut come to mind here. My preferred way is to select a lower value and add a series pot and adjust for the sweet spot, if a preferred value is close by measure and fit.
Valve amps though as has been said do not require such close tolerances, remember too the wide tolerance of capacitors, so would the O/P search for close tolerance capacitors too? I personally think that we go too far in some instances to gain only what can be seen on expensive test equipment that make no difference to the pleasure of what we are listening to.
Most of my resistor stock is good quality but 40 odd years old, very rarely do any of them read out of spec and most are 5% tolerance, they also fit in better with vintage items due to being bigger than modern types.

[Valvepower]

I regularly use Welwyn MFR5 1% resistors in Leak TL/ST et al. restorations and they work absolutely fine. I use them as I've got them in the workshop.

I was able to ‘cadge’ some 91K MFR5’s for the phase splitter load resistor.

Terry

[Luxman1050]

Thanks guys alot of info to digest there.

[Diabolical Artificer]

If you do change them it may be a good idea to a little bit at a time, rather than a wholesale job. Change a few, power up & test, discharge any caps then do the next bit. That way it's easier to find any introduced faults.

Andy.

[Luxman1050]

I'll endeavour to give it a go and see what the outcome is. Good point Andy re do a bit check do a bit more. As opposed to doing them all at once.

[kalee20]

These days, with 1% resistors being mainstream and costing pence, there's absolutely no reason not to use them.

Why would anyone want to introduce any unnecessary uncertainty into a circuit by using anything less?

For setting-up the operating conditions, it's virtually certain that the resistors will be far more certain than the valves - any differences between two examples of the same circuit, will be down to the valves. The variation due to resistors can be ignored if you use 1% components.

For those who say, you don't need that level of precision... why not? If you're making an amplifier with negative feedback, the amplifier and valve characteristics virtually disappear from the overall gain, and the final figure is determined almost exclusively by the feedback. So if you're making a precision measuring instrument, you're going to need that sort of accuracy anyway.

Me, I tend to standardise on Welwyn MFR5 and MFR4 metal-film resistors (I have to declare an interest - I work for a TT-owned company, though my preference pre-dates that era).

[PYE 405]

With modern metal film resistors, along with the power rating of course, you need to be mindful of their voltage rating when using in a valve amplifier.

[Luxman1050]

Must admit that's one area that that I'm struggling with. The voltage ratings do you go with 350v or 750v although most are 350v. Plus I've noticed 1 watt resistors seem to be thin on the ground the majority are 0.5 watt. Since it's a valve amp I'd prefer 1 watt as opposed to 0.5 watt. Then temperature coefficient most seem to sit at -55° to 155° thats fine.
I think carbon film are less noisy these days but the trend seems to be metal film. So which is best?

[Craig Sawyers]

Temperature coefficient is one. Metal film resistors are often +/-50ppm per C, and rarely worse than +/-100ppm.

Carbon film are much greater, 300ppm generally but can be worse. Always negative. So a 50C temperature rise will decrease the resistance of a carbon film resistor by 1.5%

1W metal film resistors are easily available, see for example https://uk.farnell.com/vishay/mbe041...ilm/dp/3547088

+/-1%, 500V, 1W and +/-50ppm/C

Craig