>Sure I can find a few, but they are mostly out of stock and very
>expensive. I can get 1% metal film through hole 10M?s for almost
>I think I may have been ruined by 0.1% resistors. The latest
>instrument has several decades of resistors, 10 ohms to 100M. During
>testing you can get addicted to seeing three 9?s or 0?s, then 1.008,
>?Why that?s off by almost 1%", I say to myself. And then I remember
>it?s the 1% 10Meg.
I think everyone in Beaverton, Oregon must have a laser trimmer in their backyard and a pile of alumina substrates.
I think ESI started the mess. Not sure if they still do this kind of trimming, though:
You could just buy a batch of 10M from your supplier and select out the ones you need. Of course, your supplier might have already done that and sent you what's left -- a bimodal distribution of part values without any you want. ;)
Sure I can find a few, but they are mostly out of stock and very expensive. I can get 1% metal film through hole 10M?s for almost nothing.
** When a component is not readily available or is too expensive - you have to use a work-around. This means applying a bit of grey matter.
About 10 years ago I decided to build my own 4.5 digit AC and DC voltmeter using a 20,000 count Datel LED panel meter as the heart.
The look and easy readability of a bright red LED display is way above anything else IMO.
The 4 step input attenuator is a series chain of 1% resistors and 3 x 10 turn cermet trimmers ( totalling a bit over 11 Mohms) - bypassed with close tolerance caps to make the response flat to over 500 kHz.
The chain goes: 3.3M, 3.3M, 3.3M, 220k, 50k trim, 1M, 2 k trim, 100k, 500R trim, 11k. The output points are the wipers of each trim pot and the top of the chain. Took me only a few minutes to set the 3 trims to get within 0.1 % of the needed 10:1 ratios.
The AC ranges are engaged by a relay that brings in an AD536 true RMS to DC converter and the meter is either AC or direct coupled with accuracy better than 1% out to 100kHz on all ranges.
We just barely managed to find some 50M 1% metal film resistors. They're RN55 (1/4 axial) size, so we thru-hole mounted them (at a 45 degree rotation angle!) flying above a bunch of surface-mount parts.
The "PH200", I love it. I have dreams of turning our optical pumping apparatus into a magnetometer. It needs about a 3 MHz bandwidth with ~10 uA of photocurrent (onto a 0.25" diameter PD). Unfortunately this addition might be expected to sell an additional (1) optical pumping apparatus per year. (and that's on a good year.)
Then there are lots of fun things to do with a small low noise ~
Will the PH200 allow you to trade 'gain' for bandwidth?
We are considering doing an o/e converter with two electrical outputs, one the "normal" X1 gain, and a second hard-clipped X100 or X1000 gain. That would be for lidar-ish systems where a big transmit pulse would be followed by some weak echoes, or equivalent. The one we have in mind, for a laser application, would in fact be 100 MHz bw.
Sorry for the ego problem. (Most of the time I feel pretty ignorant when reading threads on SED.) I try to let people know when I'm guessing and when I 'think' I really know something.
Can anyone here help me figure out what crucial fact about resistor tolerance I am apparently unaware of? (I know this is a bit like trying to guess what color shirt Phil is wearing.... it would be nice if Phil would enlighten me... at least a hint.)
It's not to do with a bimodal distribution, nor the tempco. So perhaps the voltage rating, aging? Are there ever any non-linear terms...
I googled resistor tolerance.. but found nothing of much help on the first ten hit's.
As John said, the wringout is done, and once the datasheet is finished, one of us will post a link.
For anyone curious, the PH200 is a moderate-bandwidth transimpedance amp that uses what the marketing department calls a 'photon coupled transimpedance architecture'. That gets it more than an order of magnitude improvement in the noise vs bandwidth tradeoff compared to other commercial TIAs, and to any of my published ones. Here's a link to the datasheet plot of noise vs photocurrent, from data taken last week by yours truly:
It's shot noise limited above about 60 nA with a 1 MHz bandwidth, and its noise is asymptotically 1.76 dB above shot noise at higher photocurrents. (1.76 dB is 10*log(1.5), which gives a hint at how it works.)
You can get noise of that order with really small photodiodes, but this one is a nice big 7 sq mm (2.3 x 2.3 mm), so you can actually get your beam on there with no big problems, and don't have to worry about focusing a 20 nW beam!
The PH200 design has benefited greatly from collaboration with John and one of his engineers, Jonathan Dufour. More stuff Real Soon Now.
The series is intended to make it easier for people to do their actual scientific work, by taking as many as possible of the nasty practical problems and nailing them to the floor, so it's like they were never there. Stuff like TIA noise peaking, laser intensity noise, ground loops, mode hopping, and so on and so forth.
The motto of the series, coined by Rob Gaddi, an occasional SED contributor, is "Be Done, Now."
(The name is an ironic reference to Hewlett-Packard's first product, of course.)