Which 8.5digit Multimeter should I get?

[maxwell3e10]

I can test it against the 3458A as soon as the meter is back against the 3458A.
I assume that the 8508A and the 8588A uses some kind of transimpedance amplifier for the small ranges. That would be a benefit, because the voltage above the feedback resistor can be quite high.

In one of their 8508A whitepaper they wrote:
"By employing appropriate design techniques, the input burden voltage that the multimeter presents at its input terminals when measuring current can be essentially isolated from the voltage developed across the internal current shunts, reducing the disturbance (burden voltage) that it presents to the source of current being measured."

It would be great to get some data on 8588!

I am not sure what they mean exactly by the above statement on the burden voltage. The spec. sheet simply lists the input impedance. I doubt it  would use a transimpedance amplifier.

[e61_phil]

The spec. sheet simply lists the input impedance. I doubt it  would use a transimpedance amplifier.

It lists the input impedance to calculate your burden. The impedance is not only the shunt normally, but also some protection.

Even if the 100Ohm for the input would be the shunt only, you need to beat the 100mV spec to achieve the DCI spec.

I found a slide from a 8508A presentation and at least the 8508A uses a transimpedance input. Therefore, I'm quite sure it is the same with the 8588A.

[Kleinstein]

There are other DMMs (e.g. DMM7510, SDM3055) that combine classical shunts for the higher currents and a transimpedance amplifier for lower currents (e.g. < 10 mA).
The transimedance amplifier may still need some series impedance and thus may not have a burden that is ideally zero.

[maxwell3e10]

In 8508 the input impedance keeps increasing by a factor of 10 with each lower current range and the lowest range is only 200 uA. In 8588, there is an extra range of 10 (20) uA that keeps the same input impedance (100 Ohm), but has 10 times higher resolution.  So that probably does use a transimpedance circuit.

[e61_phil]

In 8508 the input impedance keeps increasing by a factor of 10 with each lower current range and the lowest range is only 200 uA. In 8588, there is an extra range of 10 (20) uA that keeps the same input impedance (100 Ohm), but has 10 times higher resolution.  So that probably does use a transimpedance circuit.

And why not all small ranges? Bringing all the circuit into the instrument just for using it with the lowest range wouldn't make any sense to me. Especially not, if the predecessor already used that technique for higher ranges. In my opinion it is much more likely, that they are using a post amplifier for the lower range and therefore, the same input impedance.

[Kleinstein]

With a TIA one can use the some resistor for more than one decade. It would totally make sense to use them for 2 maybe even 3 ranges and between those ranges only change the votlage gain. So the lowest range may use the TIA with something like 1 V maximum output and the next range would use the same TIA setting but up to 10 V.
Using a TIA for higher currents also comes with a downside: with a higher voltage at the shunt / measurement resistor the self heating becomes an issue. So for somerhing like 100 mA there is a good reason the limit the voltage at the shunt to some 200 mV. With much more drop one could get less noise, but also starts to get INL problems from heating the resistor. For the low currents it is also relatively simple to have a TIA: it is mainly a single (it may need to if a 5 V supply type is used) precision op-amp (maybe a zero drift one) and the resistor. It may even be easier to use completely separate TIAs for different ranges insteat of range switching around a single amplifier. The higher currents would need an extra power stage.

With the way the higher end Fluke meters implement the 4 wire ohms mode (with the reference resistor from the ground side), one may reuse part of the circuit for both the TIA and current source. So depite of the extra effort and still limited voltage drop it may still make sense to have a TIA also up to 100 mA (as the ohms test current).

[Fidel]

A few prices for the Fluke meters were mentioned, if anyone is interested the US retail pricing as of May 1,2022 the 8588A is $18,553 and 8558A $12,789
The 20min transfer specs are almost identical, 10V DC is identical - so if you have your own standards to compare to the 8558A option would be a great option.

[binary01]

I'm sure I cannot add much of technical value to the discussion, compared to other experts here, but just commenting that I opted for the 8588A in my small accredited lab, and I find them to be excellent from a useability and breadth of capability perspective.
I actually purchased two of them, with external calibrations occuring in alternating cycles so that I can perform cross-checks and continue running with one while the other is away for calibration.  I only have a couple of years history (back to Fluke UK each time), so it's hard to draw strong conclusions on stability, but so far the stability on all functions has been good against the 1 year specification, so I am confident that true performance is well within my uncertainty budgets.

I have used and had always intended to purchase the much-loved 3458A when developing my electrical capability, but in the end I assessed that the configuration of the 8588A ranges and extra capabilities might be better for my application.  I certainly have no complaints from my experience with the 3458A, and although not well founded in real data, my understanding is that the 3458A might be the ultimate instrument if you are very strongly focussed on DCV linearity. It's difficult to get specific linearity specs for the 8588A to compare, but in my case the absolute accuracy spec of the 8588A DCV (including non-linearity, temperature and 1 year drift) was still substantially better than I required. 

I don't usually use the 30A range on the 8588A (I prefer to use an external shunt instead), however one of the well regarding reference labs in Australia has commented to me that the 30A range in the 8588A is significantly better than the 20A range in the 8508A in their experience, most likely due to the addition of active cooling. 
Although it's only simple, I do like the external shunt feature on the 8588A - I have them programmed with all of my DC shunts and find it very handy to take readings directly in current units.
A dissapointment for me is the limited range (with 10% only) in which the transfer spec is used for ratio measurements.  This has already been discussed here and in other threads, and is a significant impediment to using the 8588A to transfer traceability between standards of different values (i.e. across decades) and perhaps hints at a lower confidence from Fluke in the linearity over a full range.

[bdunham7]

And why not all small ranges? Bringing all the circuit into the instrument just for using it with the lowest range wouldn't make any sense to me. Especially not, if the predecessor already used that technique for higher ranges. In my opinion it is much more likely, that they are using a post amplifier for the lower range and therefore, the same input impedance.

I would be curious to know exactly how they've implemented this and maybe why.  Fluke has used a variety of methods on other products. They used post-shunt 10X gain on the 8842A because they had added it for volts and ohms.  The 8845/6A use a digital gain extension that increases noise and worsens the accuracy spec, so hopefully they didn't do that with the 8588A.

Their 850x meters have used a TIA for the lower current ranges starting from the very beginning with the ancient 8500A.  And this works for both DC and AC current measurements, the output of the current shunt module is just passed to whichever AC converter is installed. 

Here's an explanation from the 8500-series manual.  The end result is that the input burden for the lower three current ranges is 10R + ~2-3mV.  Unless there is some other issue, this seems to be an optimal solution.

[Oldtestgear]

Not sure if this is relevant or helpful but here goes. I bought an HP 3458A in September 2021 from a trader (not  a TE dealer). From the Keysight history it was last calibrated in 2012. When sent for calibration in November it was still in specification with no need for adjustment. In fact they calibrated it twice in as many weeks (please do not ask!) with the same result.  This was after I had replaced the PSU caps, mains filter + associated caps & the NVRAMs. I was careful to save the calibration constants before replacing the NVRAM.  The NVRAM may well have been the original one from new.

Suggests to me that an older 3458A may still be a good buy for a cal lab where stability must be a major consideration.

Yes the design is 30+ years old with a bit of reworking for obsolescence issues  but that does not make it a bad option. To be fair I have no experience of the Fluke but would be happyto try it if one comes my way for a decnt price. Not holding my breath on that one.

FWIW

Phil

[mendip_discovery]

Part of the decision to go new was second hand prices are 6k and new its 9k. The ones the MD saw were in a tired looking state so thought it might be better to get new. I am still waiting on quotes from places. Clearly, they are in no hurry to sell gear.

[boggis the cat]

Also worth considering that the Fluke is manufactured in the UK, and calibration by Fluke is carried out there.  So they would be local suppliers and service providers.

No significant difference for my work as we are located in New Zealand, but as you are in the UK this could be a big advantage.

As to the differences between these two instruments - the Fluke is far more capable, and much easier to use.  We have a couple of 3458As, an 8508A and the 8588A in our lab.

Ensure that any 'options' are included in quotes.  These can be significant extras.

[binary01]

No significant difference for my work as we are located in New Zealand, but as you are in the UK this could be a big advantage.

I agree, OP is lucky to be so close to Fluke UK!  The freight alone approaches AUD $1000 for my 8588A to take a return holiday to Norwich UK (Fluke) from Melbourne.  Tempting to fly there myself and take it in my luggage next time 

[RYcal]

Also worth considering that the Fluke is manufactured in the UK, and calibration by Fluke is carried out there.  So they would be local suppliers and service providers.

No significant difference for my work as we are located in New Zealand, but as you are in the UK this could be a big advantage.

As to the differences between these two instruments - the Fluke is far more capable, and much easier to use.  We have a couple of 3458As, an 8508A and the 8588A in our lab.

Ensure that any 'options' are included in quotes.  These can be significant extras.

You now have my attention...Can I ask what lab you are in, PM me if you like