dogears

Eliminating trimmers: John Bat Signal
« on: December 09, 2020, 10:08:17 AM »
Hey John, something you've mentioned several times was that Peavy did everything possible to eliminate trimmers.

Can you expand on how you'd do that from a design perspective? Was it primarily by component tolerance sorting / binning, or custom component ordering, or...?


abbey road d enfer

Re: Eliminating trimmers: John Bat Signal
« Reply #1 on: December 09, 2020, 12:52:58 PM »
Hey John, something you've mentioned several times was that Peavy did everything possible to eliminate trimmers.

Can you expand on how you'd do that from a design perspective? Was it primarily by component tolerance sorting / binning, or custom component ordering, or...?
One has to know what performance really matters. Gain accuracy? Most of the times gain is set by ear, so about +/-3dB accuracy is all that's required. That's actually about -30/+40%. using 5% resistors will give you that easily.
But actually, in order to maintain a proper level diagram, each stage needs to be better controlled. 5% resistors will guarantee less than 1dB deviation. That is usually considered inadequate, so using 2% or 1% tolerance for resistors is quite common.
CMRR. That's where you want matching, not necessarily precision/accuracy. A DIY'er may use 10% resistors and match them with a multimeter. Industry will purchase 1% or better resistors.
Capacitors. Electrolytics are notouriously inaccurate, typically +/-20% or worse, so designers overspec, put 100uF where calculation says 10uF.
Those that are used in EQ's need to be more accurate. For tone controls, 10% tolerance is good enough (who can hear the difference in a shlf EQ at 90Hz instead of 100?). Bell eq's are a little more demanding, so may require 5% tolerance.
The main area where audio designers are concerned with reproductibility is transistors, particularly power or low-noise devices; manufacturers work together with suppliers for specially graded devices.
The rest is up to the designers, making sure that the performance is not unduly dependant on the parts accuracy. Generally, making sure the circuit operates under strong Negative FeedBack is sufficient.
« Last Edit: December 09, 2020, 12:55:59 PM by abbey road d enfer »
Who's right or wrong is irrelevant. What matters is what's right or wrong.
Star ground is for electricians.

buildafriend

Re: Eliminating trimmers: John Bat Signal
« Reply #2 on: December 09, 2020, 02:50:00 PM »
In my commercial electronics experience being the first tech that got hit with devices before it went to 2nd and 3rd QC checks they remove these devices for many reasons. One of them is that then you have less to calibrate and another is that you don't have a trimmer for someone else to to try to figure out the purpose of. Sometimes there are ways to stabilize the biasing of a circuit without the need for a trimmer. It's also another thing that can break or wear out. personally, I love trimmers but I also believe we don't want to be giving people more to figure out. I bet higher level people in here will have other and stronger points

JohnRoberts

Re: Eliminating trimmers: John Bat Signal
« Reply #3 on: December 09, 2020, 02:52:33 PM »
Hey John, something you've mentioned several times was that Peavy did everything possible to eliminate trimmers.

Can you expand on how you'd do that from a design perspective? Was it primarily by component tolerance sorting / binning, or custom component ordering, or...?
Design... and tight component specifications.

For the example I mentioned of biasing up class A/B amplifier output stages, at the time Peavey was probably Motorola's largest customer for TO-3 power transistors and we had several complementary pairs in our system tightly specified to behave well at popular power points (Peavey sold lots of amplifiers). The double diode, also purchased from Motorola was tightly specified to perform well with the preferred Peavey output stage pairs. I don't know all the details of that design process because it was done before my time there (mid-80s). I felt little reason to re-invent the wheel for amplifier stages inside my powered mixers.

The trim-less class A bias was a bit of a compromise, amplifiers when cold could be a little bias starved (more on that later) but across millions of amplifier channels it was proved robust and reliable. At one point when I was working as product manager over all Peavey power amps, I inherited a heavy iron class A/B amp (PV1.3k). These amps had a less than stellar reputation for thermal stability. This is a little inside baseball about amplifier design but the susceptibility for thermal runaway is a function of thermal resistance junction to ambient.... In other words, amplifiers with marginal power dissipation capability are more likely to run-away thermally and melt down. Peavey was a sharp pencil company when it came to approving cost increases to existing products, but as product manager with my awareness of how amplifier worked, I was able to negotiate the cost increase in a new product authorization. I rebranded the PV1.3k as the PV2000, pretty much the same amp but with 4 more TO-3 power transistors. I couldn't add heatsink, but adding more power devices reduced the effective thermal resistance junction to ambient making the amp rock solid thermally. The extra devices also lowered the distortion because the individual power devices weren't working as hard but that was just gravy.
=====
For the "more" about starved class A bias when cold, Peavey had a modest power guitar amp that used plastic darlington output devices. That one amp model was notorious for sounding ratty on cold winter mornings if the dealer didn't warm it up before demoing it to a potential customer.

Wearing my design engineer hat, I actually came up with a slick (IMO) self biasing scheme using a LTP detecting the voltage drop across the emitter degeneration resistors to crudely regulate class A bias current irrespective of the power devices. When I told Jack Sondermeyer I could fix his amp, he said "show me"... so I did. My modified guitar amp worked but that was not enough to get Jack to redesign one of his amps that was already in production.

For me to prove my design worked I would have to design it into one of my new products for full production. The opportunity for that was a high tech, dsp based, Karaoke top box I was tasked with overseeing. This SKU had a modest 35W power amp, perfect for proving my design, BUT..... at the time my political stature was compromised so I was forced to report to the digital engineering director (he was a much better politician than me  ::) ). He refused to gamble on using my unproved amp design that could delay the product introduction or cause service issues if it didn't work perfectly (another reason to never work for a boss). Ironically that Karaoke SKU suffered horrendous field service problems due to a Japanese cassette transport that would jam and lock up... It took me the better part of a year trying to figure out what I did wrong that was causing the lock ups. It was the same exact transport I had used in the Peavey 4T cassette but using a cheaper 2T stereo head. The Japanese vendor finally admitted that they had changed the durometer of one of the plastic gears inside, causing the lock-ups. By that time the product had earned a terrible market reputation and was irredeemable.

===

Also on the topic of designing out trims, one of my several patents while working at Peavey was for a trimmer-less VCA.

Quote
US04818951  Roberts

04/04/1989   Gain control or multiplier circuits. This was a circuit design that utilized separate DC and audio frequency feedback paths to automatically bias transistors in a VCA design so unmatched transistors could be used in production. This VCA was used in the old AMR CDS-2 compressor/limiter/de-esser. The Patent was assigned to Peavey Electronics. 

This was a useful to avoid factory trims and selected parts but class A VCA was not very high performance and eventually abandoned.


This was back before dbx/THAT corp freely shared their VCAs with potential competitors. This class A VCA was good enough for a cheap -10dBV comp/limiter/de-esser but just barely. Later when the good dbx VCAs were made available Peavey used THAT corp VCAs in dynamics processors (DOH). 

JR           
It's nice to be nice....

dogears

Re: Eliminating trimmers: John Bat Signal
« Reply #4 on: December 10, 2020, 12:09:47 AM »
Interesting all around. Thanks!


 

Related Topics

  Subject / Started by Replies Last post
26 Replies
13559 Views
Last post August 05, 2004, 11:25:14 PM
by Sleeper
6 Replies
2484 Views
Last post August 16, 2006, 02:10:23 PM
by gyraf
1 Replies
1520 Views
Last post July 12, 2007, 02:02:13 PM
by NewYorkDave
4 Replies
1721 Views
Last post January 25, 2008, 10:00:51 AM
by crazytooguy