GenRad Decade Inductance, Capacitance, and Resistance Boxes

[KungFuJosh]

Today I got in a General Radio "GenRad" 1434-QC Decade Resistance box.

I think I lucked out; all it really needed was to be cleaned and have the knobs aligned.

I ran through a cal (attached) with my IM3570 basing limits on +/- 0.02% of the maximum value of each decade. I didn't account for aging, or anything else...and it still did well. I think it's impressive for something that could be around 50 or 60 years old. Allegedly, this model was introduced 61 years ago, but the datasheet I attached is from 1974.

The design looks pretty interesting too. I'm curious how it compares to their new boxes.

Thanks,
Josh

[trobbins]

Am I misinterpreting the accuracy specs of the IM3570, but it looks like 'as new' basic accuracy spec for resistance at dc is +/-0.1% of reading + a titch ??  The original GR spec is +/-0.02% incremental for most decades.

It is intriguing to check for drift over many decades, which hopefully is not much more than original accuracy spec, but without the benefit of knowing the original tolerances.  I've yet to do that with my ESI DB62 - which doesn't appear to have gone off the rails, but I have yet to add any rigour to that check.

PS. given your IM3570 - you could possibly take an impedance scan over frequency to see when the ranges start to deviate from pure resistance.  I've even tweaked an old decade box to effectively neutralise the inductive component out to 100kHz on a few decades where the ww coils had noticable inductance.

[Martin72]

I've had this set on my watch list on eBay for ages:

GR 1423-A

But I'm kind of hesitant because of the price.
Even though it really looks great.

[KungFuJosh]

I've had this set on my watch list on eBay for ages:

GR 1423-A

But I'm kind of hesitant because of the price.
Even though it really looks great.

I remember that one. I hate to say it, but for that model and as clean as it looks, I think it's a good price. They also accept returns, so it's low risk. Anything I saw that looked close to that, was thousands instead of hundreds.

That thing is HUGE, looks pretty recent, and clean. If you want something less expensive, you can get older models with 3 or 4 decades for much less. However, many of those are sold as Parts/Repair, so they come with some risk.

BTW- I had no idea how much I'd like having this decade box. It's so much easy to work with than adjusting a tiny pot. 

[KungFuJosh]

Am I misinterpreting the accuracy specs of the IM3570, but it looks like 'as new' basic accuracy spec for resistance at dc is +/-0.1% of reading + a titch ??  The original GR spec is +/-0.02% incremental for most decades.

The claims are:

The Hioki IM3570 Impedance Analyzer features a basic impedance (Z) accuracy of ± 0.08% rdg. and typically achieves a basic DC resistance (Rdc) accuracy of ± 0.05% rdg. in its high-accuracy DC measurement modes.

I don't know of any lab grade LCR meter with better than ± 0.05% basic accuracy. If you have one you'd like to donate, please send it my way.

It is intriguing to check for drift over many decades, which hopefully is not much more than original accuracy spec, but without the benefit of knowing the original tolerances.  I've yet to do that with my ESI DB62 - which doesn't appear to have gone off the rails, but I have yet to add any rigour to that check.

The DB62 looks really nice. Please share some photos of the guts if you don't mind.

PS. given your IM3570 - you could possibly take an impedance scan over frequency to see when the ranges start to deviate from pure resistance.  I've even tweaked an old decade box to effectively neutralise the inductive component out to 100kHz on a few decades where the ww coils had noticable inductance.

I do plan on doing some sweeps eventually. Right now though the decade box is attached to my K2010 WIP.

How did you tweak the old decade box?

Thanks,
Josh

[mawyatt]

I've had this set on my watch list on eBay for ages:

GR 1423-A

But I'm kind of hesitant because of the price.
Even though it really looks great.

I remember that one. I hate to say it, but for that model and as clean as it looks, I think it's a good price. They also accept returns, so it's low risk. Anything I saw that looked close to that, was thousands instead of hundreds.

That thing is HUGE, looks pretty recent, and clean. If you want something less expensive, you can get older models with 3 or 4 decades for much less. However, many of those are sold as Parts/Repair, so they come with some risk.

BTW- I had no idea how much I'd like having this decade box. It's so much easy to work with than adjusting a tiny pot. 

That does look very nice and from GR you know it's well made!! However, we wouldn't put much faith in the 100pF lower range readings with banana jacks as the connection means!!

With a +-0.05% specification one is going to have a very difficult task achieving anything close to this specification, or repeatability, even with the best Kelvin clips and LCR meter (that's +-50fF at 100pF). Here a BNC or better direct connector/cable is in order. This is why we developed the direct BNC to LCR meter (no cable, direct BNC to BNC) type capacitive references which eliminates the Kelvin cable and clips or other fixtures. Especially effective for the low capacitive ranges which have shown excellent repeatability with "aged" Mica type capacitors.

BTW we are considering offering these direct BNC capacitive references with select Mica capacitors as kits if anyone is interested. We've done enough test/evaluations to know they are quite stable and repeatable, so comfortable to let others experience the benefits of such.

Best

[KungFuJosh]

That does look very nice and from GR you know it's well made!! However, we wouldn't put much faith in the 100pF lower range readings with banana jacks as the connection means!!

I think you'll have to get one also and compare.

It does have 4 wires + guard, so it should be reasonable...I would hope considering the price tag new.

[mawyatt]

That does look very nice and from GR you know it's well made!! However, we wouldn't put much faith in the 100pF lower range readings with banana jacks as the connection means!!

I think you'll have to get one also and compare.

It does have 4 wires + guard, so it should be reasonable...I would hope considering the price tag new.

We don't need to get one to know about the parasitic capacitive issues with using banana jacks and cables as a measurement means

You've already experienced the Gore cables and why they are the "golden standard" for RF/MW cables, and this reflects into the lower frequency ranges where these decade capacitive boxes are intended to operate. Even the best Kelvin cables/clips will experience small distributive capacitive parasitic effects (cable movement/position/orientation/surroundings as such), which will show in 100pF and below ranges at these precision levels.

Remember that parts of the measurement are "outside" the guard shielding with Kelvin Clips, and why Hioki recommends a "guard shield plate" below the Kelvin measurements, this significantly reduces parasitic field effects, but doesn't totally eliminate such, and can't help with Kelvin cable movement induced effects. For DCR measurements most of these effects aren't an issue, same for much larger capacitive (lower Z) measurements where outside parasitic effects are swamped by the low Z DUT altho the parasitic series effects (inductance and resistance) now come into play.

So basically there's no free lunch with precision capacitive measurements, small caps require different techniques/approaches than high capacitive measurements, same for resistors and inductors!!

Best

[KungFuJosh]

I think the guard connecting the metal box instead of a shield plate should be sufficient in that reguard.

I do know all that, but I'd still like to see for myself how it measures. Theory aside, for a device that expensive, from a company that well respected, I'm at least cautiously optimistic that the range should be at least a little bit usable.

There's also to consider the absolute accuracy/uncertainty vs how important that is for a specific use case.

[zepto]

That does look very nice and from GR you know it's well made!! However, we wouldn't put much faith in the 100pF lower range readings with banana jacks as the connection means!!

With a +-0.05% specification one is going to have a very difficult task achieving anything close to this specification, or repeatability, even with the best Kelvin clips and LCR meter (that's +-50fF at 100pF). Here a BNC or better direct connector/cable is in order. This is why we developed the direct BNC to LCR meter (no cable, direct BNC to BNC) type capacitive references which eliminates the Kelvin cable and clips or other fixtures. Especially effective for the low capacitive ranges which have shown excellent repeatability with "aged" Mica type capacitors.


Best

The 100pF decade is 0.15% + 50fF. The minimum step setting is 100pF to account for residual. It has 2 and 3 terminal with a specified difference. I don't think the spec is unreasonable, but as you say hard to verify with typical LCR meter.

[mawyatt]

I think the guard connecting the metal box instead of a shield plate should be sufficient in that reguard.

I do know all that, but I'd still like to see for myself how it measures. Theory aside, for a device that expensive, from a company that well respected, I'm at least cautiously optimistic that the range should be at least a little bit usable.

There's also to consider the absolute accuracy/uncertainty vs how important that is for a specific use case.

Yes, the meter box is guarded, however the banana connectors are outside the guard and anything that connects to them is also outside the guard. The two box mounted banana connectors alone with nothing connected likely contribute more than 50fF.

Banana connectors are not a good choice for a low valued capacitive reference IMO, others may have a different opinion tho.

Best

[KungFuJosh]

If nothing else, your opinion has opened me up to looking at 3-decade boxes that exclude that range.

[Martin72]

How far can you go with parasitic capacitance compensation? Or, to put it another way, if I set the decade range to 1 pF and then had to feed that value to my Hioki IM3536-CE via a suitable test lead, would I still measure 1 pF?

[mawyatt]

That does look very nice and from GR you know it's well made!! However, we wouldn't put much faith in the 100pF lower range readings with banana jacks as the connection means!!

With a +-0.05% specification one is going to have a very difficult task achieving anything close to this specification, or repeatability, even with the best Kelvin clips and LCR meter (that's +-50fF at 100pF). Here a BNC or better direct connector/cable is in order. This is why we developed the direct BNC to LCR meter (no cable, direct BNC to BNC) type capacitive references which eliminates the Kelvin cable and clips or other fixtures. Especially effective for the low capacitive ranges which have shown excellent repeatability with "aged" Mica type capacitors.


Best

The 100pF decade is 0.15% + 50fF. The minimum step setting is 100pF to account for residual. It has 2 and 3 terminal with a specified difference. I don't think the spec is unreasonable, but as you say hard to verify with typical LCR meter.

That's a little better spec-wise, but still quite optimistic IMO. I'm sure GR can achieve this with an OEM dedicated test fixture setup, however others may have some difficulty unless they can precisely duplicate the OEM test setup and environment.

We've found with LCR meter direct connected BNC wired fixtures like the precision Tonghui TH26048A which the DUT component directly plugs into (for wire-based components) and the fixture directly mounted to the LCR Meter via direct BNC connections, this level of precision/repeatability needs close attention to details. Just clipping on a couple Kelvin Clips to a pair of open air banana connectors will be difficult to achieve this level of precision and repeatability with 100pF cap reference IMO. I'm sure one can do this with the right amount of "fudging" like we've seen with some tweezer measurements, but not with a random "blind" measurement (example here's an unknown precision cap).

Anyway, these are just our experiences and opinions, as always YMMV.

Best

[mawyatt]

How far can you go with parasitic capacitance compensation? Or, to put it another way, if I set the decade range to 1 pF and then had to feed that value to my Hioki IM3536-CE via a suitable test lead, would I still measure 1 pF?

Don't think that's possible at 1pF to the level of precision/repeatability being discussed. This is typically VNA with dedicated test fixture territory.

One can certainly read with good results 1pF with a quality LCR meter (like IM3536), but requires very precise and stable (both electrical and mechanical) test fixturing, certainly not banana connections!! We've done this level of capacitive measurement often with the modified Plunger type direct BNC connected LCR meter fixture (no external cables involved) for high Z SMD devices. Also at this level we have an aged 2pF Mica reference capacitor (a 10pF, and more) that are mounted in our custom direct BNC enclosures. These enclosures provide a decent mechanical and electrical environment for the small reference cap, and have shown very good repeatability over multiple frequencies and measurement sessions.

BTW if you followed our post related to these LCR meter reference capacitors and enclosures, we also included the ability to measure the DUT temperature since a direct LCR meter connection has the drawback that the meter BNC connectors conduct heat towards the attachment. We found this causes 1.6~2C rise in DUT temperature on TH2830 and IM3536 LCR meters, which takes 20~30 minutes to stabilize. The DUT capacitors are directly soldered to the 4 BNC connectors (Hc, Hp, Lp, Lc) to minimize parasitics effects, with the Hp and Lp sense BNC terminals directly across the DUT which has the wires extending to the Hc and Lc terminals. This approach creates a direct 4 wire Kelvin connection to the DUT with the sensing closest to the DUT body.

Edit: Should also mention this direct type precision LCR DUT fixture provides a convenient means to establish a precise Open and Short calibration, which is required for repeatable LCR measurements, similar to the Short Open of Short Open Load Thru precision cal kits for VNA use. Here's some examples of these enclosed reference capacitors.

Best

[Martin72]

Here's some examples of these enclosed reference capacitors.

 

Excellent, really....

[KungFuJosh]

Here's some examples of these enclosed reference capacitors.

 

Excellent, really....

Yeah. Now we need to convince Mike to make some for us.

[KungFuJosh]

...and now I ordered a GenRad 1419-K. 

[mawyatt]

Here's some examples of these enclosed reference capacitors.

 

Excellent, really....

Yeah. Now we need to convince Mike to make some for us.

If we can get enough folks interested we'll make up some kits for the Open, Short and some "aged" Mica values like 10, 100, 1000pF values if these are still available. The old Soviet Micas are exceptionally good, with very low DF and good thermal stability, but are hard to come by these days.

Here's more details:

https://www.eevblog.com/forum/projects/direct-lcr-meter-connect-test-subjects/

Best 

[KungFuJosh]

Dooooo it

[mawyatt]

We'll need about 10~20 kits to make this happen. Need to get the not very common male pin panel BNC connectors (4 per), and Mica Caps (need to hand test and select from a batch), brass inserts (8 per), SS bolts (8 per), 2~3 rolls of Copper foil, thick Copper sheet (for Short), spool of heavy magnet wire, 2~3 spools of 3D filament, thermistors and connector (if used) and so on.

Patience pays off as these take awhile to assemble, and why we are considering them as kits, no way we are going to assemble them ourselves. Will print up the 3D Connectors Lever Arms, the Covers and Case, insert the brass threaded inserts and leave the rest as a kit for assembly for those that are interesting. Costs will be reasonable but not dirt cheap, the 4 special BNC connectors alone set the starting point!!

The end result is one is rewarded with an extremely stable and repeatable reference capacitance

Best   

[KungFuJosh]

If/when you figure out the pricing, I'm sure that will help get an idea of how many people want some kits.

[mawyatt]

My seat of the pants guess is about $30~40 give or take per reference cap fixture. Getting the components in enough quantity helps reduce the shipping costs & tariffs (multiple sources), so the more kits the lower the overall costs.

We've found Mica caps the most stable, especially the old Soviet (if you can find them) types. The ones we have are in metal tubs with soldered seals (hermetic), and are quite stable. In these Direct BNC Connect test fixtures we are talking 5~6 decimal places repeatability for the best ones, of course this also takes advantage of the Direct BNC Connect Short and Open Calibration

Best

[trobbins]

Wrt the 1434 box, it is only spec'd for incremental steps within each decade, not for absolute accuracy, so eg. +/-0.02% for 1k steps (or +/-0.2R per step).  Your measurement sheet indicates a worst-case step difference of 0.52R (between 8k and 9k step), which may be degraded by your meter's linearity.

I don't see your typical 0.05% accuracy in this linked datasheet: https://idm-instrumentos.es/wp-content/uploads/2016/02/IM3570E7-53E-IDM.pdf - is that your meter's datasheet?

DB62 internals (not my photo and I can't locate its origin).  It's long term (2 year !) incremental step spec is also 0.02% for 1k steps.  I'm hoping to soon have a confident measurement setup to check the step tolerances, and am expecting to see noticeable drift over the many decades of life - such is life for resistors.

The impedance compensation of the vintage decade box is indicated in https://www.dalmura.com.au/static/GAMBRELL%20BROS%204x%20decade%20resistance%20box.pdf, but it doesn't show a photo of the added caps between the coil tops.  I use the box for speaker load resistance measurements (at low power for valve amps) where I'm happy that the load is effectively resistive out to 100kHz.

Imho, the hassle with primary standard caps (I have a GR1404-A 1nF) is the repeatability of the connection interface to the measurement instrument being checked.  My goto LCR is a MCP BR2822 with kelvin clips, which at 1kHz, direct to inner terminals of 1404 measures 1009.6 +/-0.3pF, 0.0002 +/-0.1 D, and rising to 1012pF with LCR meter away from 1404.  To use the 1404 as a periodic check of the LCR meter, I could continue to just do kelvin clips to the inner part of the GR874 sockets, but that seems prone to a variation similar to the 0.3% basic tolerance of the meter.  I'd likely have to set up a custom 4-terminal breakout from the LCR front panel to a set of GR874 locking adaptors.  Given the construction quality of the GR1404, and that it has had one cal indicating +11ppm +/-20ppm back in 1989, but with no initial factory cert, and a spec'd drift of below 20ppm/yr, I'd just be hoping it was still within say 100ppm (0.01%).

[KungFuJosh]

Wrt the 1434 box, it is only spec'd for incremental steps within each decade, not for absolute accuracy, so eg. +/-0.02% for 1k steps (or +/-0.2R per step).  Your measurement sheet indicates a worst-case step difference of 0.52R (between 8k and 9k step), which may be degraded by your meter's linearity.

Absolute accuracy vs. 0 is spec'd in the paragraph above the incremental accuracy.

I don't see your typical 0.05% accuracy in this linked datasheet: https://idm-instrumentos.es/wp-content/uploads/2016/02/IM3570E7-53E-IDM.pdf - is that your meter's datasheet?

That is the datasheet, and it's weird. If you search it, you will see general is 0.08% for that model. There're more detailed accuracy specs if you want to do math.

DB62 internals (not my photo and I can't locate its origin).  It's long term (2 year !) incremental step spec is also 0.02% for 1k steps.  I'm hoping to soon have a confident measurement setup to check the step tolerances, and am expecting to see noticeable drift over the many decades of life - such is life for resistors.

The impedance compensation of the vintage decade box is indicated in https://www.dalmura.com.au/static/GAMBRELL%20BROS%204x%20decade%20resistance%20box.pdf, but it doesn't show a photo of the added caps between the coil tops.  I use the box for speaker load resistance measurements (at low power for valve amps) where I'm happy that the load is effectively resistive out to 100kHz.

Imho, the hassle with primary standard caps (I have a GR1404-A 1nF) is the repeatability of the connection interface to the measurement instrument being checked.  My goto LCR is a MCP BR2822 with kelvin clips, which at 1kHz, direct to inner terminals of 1404 measures 1009.6 +/-0.3pF, 0.0002 +/-0.1 D, and rising to 1012pF with LCR meter away from 1404.  To use the 1404 as a periodic check of the LCR meter, I could continue to just do kelvin clips to the inner part of the GR874 sockets, but that seems prone to a variation similar to the 0.3% basic tolerance of the meter.  I'd likely have to set up a custom 4-terminal breakout from the LCR front panel to a set of GR874 locking adaptors.  Given the construction quality of the GR1404, and that it has had one cal indicating +11ppm +/-20ppm back in 1989, but with no initial factory cert, and a spec'd drift of below 20ppm/yr, I'd just be hoping it was still within say 100ppm (0.01%).

Thanks for that link.

Connection quality always matters. I'm looking forward to screwing around with the decade capacitor when it arrives.