EEVblog #1005 - Keithley 2302 Battery Simulator Teardown

[EEVblog]

What's inside the Keithley 2302 Battery Simulator
A way to simulate battery ionic resistance (ESR) and measure product pulse power consumption.

[plesa]

Thanks for video. There are multiple similar block with mk1 2400 series. Also the Densei - Lambda PSU are used in more recent 26xx designs. The 2400 series use cheaper one skynetpower.com.tw.

[tszaboo]

The power supply also has a digital voltmeter (DVM) that is independent of the power supply
circuit. The DVM can measure up to +30V (1mV resolution).

So the separate section with the slope converter is probably this.

The Model 2306 power supply (dual channel battery/charger simulator — see Figure 1-1) can
simulate a battery (Channel #1) or a charger (Channel #2). Figure 1-2 shows the Model 2306-VS
front and rear panels.

The second channel's operation seems to be very different than the first channel?
Source: http://www.tek.com/dc-power-supply/series-2300-battery-simulating-dc-power-supplies-manual/models-2302-2302-pj-2306-230
Models 2302/2302-PJ/2306/2306-PJ/2306-VS Battery/Charger Simulator Instruction Manual

Its refreshing to see something new test gear.

[alm]

I like the use of high-res photos instead of live video for the detailed parts of the details.

What I find interesting is that the 2302/2306 battery simulators use a display with much better contrast (VFD?) when new than the closely related 2303/2304 high speed power supplies. Cost cutting? Were they worried about aging? They have the same terrible controls, however.

So the separate section with the slope converter is probably this.

I imagine the ADC handles both the readback and the separate DVM channel. Note that the DVM channel has to stay between -5 V and +30 V from the negative output terminal. In the thread about the 2304A I linked below, TiN says that they report the full 6.5 digit value via GPIB. Accuracy will obviously be worse than a true 6.5 digit DMM due to the voltage reference.

The second channel's operation seems to be very different than the first channel?

The specs are a bit worse for the second channel and it lacks the variable output impedance, probably because it is intended to simulate a charger (which might be a plain SMPS power brick in real life). Not sure what you mean by very different operation? From the front panel they work exactly the same, and I would expect the circuit to be very similar.

What I like over these devices over the Agilent battery simulators from the same era is that they let you program the sink current. So it can function as a true programmable two-quadrant power supply. Maximum current is quite limited however, especially at higher voltages. This unit was clearly optimized for ~4 V operation. The half-rack Agilent units will have a fixed maximum sink current.

They could (at least before this video came along ) be found relatively cheaply on eBay (around $200). Probably a bunch of cell phone design places that dumped them for newer models.

[kandrey89]

I just bought Keithley 2304A for $125, it works, had to spend 2-3 hours taking it apart and using a brush a dusters to get all the dust out.

What's the connector model number for the back, mine is missing, so I need one so I can connect wires to the instrument?

[pigrew]

This instrument seems very comparable to the Agilent 66311D and 66321D instruments, although I find the Agilent ones much easier to work with.

I've tried to like it, but I'm not able to think of any advantage of it over the Agilent units. Perhaps it has better transient response?

Its SCPI interface much less usable for my development work. For example, to produce an oscilloscope-like view of current, it must be triggered on a current pulse (at least with the 2304A, no immediate trigger is available). While it advertises amazing 33us integration times during digitization, every acquisition has to be followed by 241us of processing time. At least they've implemented over-voltage protection on the 2302/2306 (which the 2303/2304 lack).

Anyway, I've made a 3D-printable bench kit with legs that match middle-aged Agilent equipment, for those of you that got the rack-mount version.

[pigrew]

What's the connector model number for the back, mine is missing, so I need one so I can connect wires to the instrument?

I'm using Phoenix Contact 1757077 with the Keithley 2304A. It doesn't have the side-screws, but it is quite snug.

[alm]

I've tried to like it, but I'm not able to think of any advantage of it over the Agilent units. Perhaps it has better transient response?

In the case of the 2304A (unlike the 2302), it has a 20V / 5A range, which is wider than that of the Agilent Communication supplies (15V / 3A, burst up to 5A). I do not think that any of the 661xC series (i.e. the precision bench power supply version) go that high either. The Keithley 2302 and 2306 are generally cheaper than the Agilent 66319/321D series. Both the Keithley 2302/6 and the newer Agilent supplies have some extra features compared to the older 66309/66311B units (DVM, variable output impedance, voltage limit).

The main advantage in my opinion is the programmable sinking current. The Agilent supplies (at least the half-rack ones) will sink up to some fixed limit (depending on the voltage). You need to step up to the full rack width units to get this feature in Agilent supplies.

Apart from the advantages I mentioned, the Agilent communication supplies are better units with a much nicer front panel in my opinion.

[bushjavier]

@pigrew I agree with you i have the 66311D, i use it for my development process.

also few minutes ago i posted this

https://www.eevblog.com/forum/testgear/agilent-66311d-and-the-agilent-an-1427-application-note/

Do you think is possible to use the 66311D to read current as described in the an 1427 app note?

[tec5c]

I would be curious to know the output noise of an instrument like this.

As Dave has done a video on low frequency noise measurements of an op-amp, in which he correctly stated (and used) that batteries are often used as a bias supply to rid of any extra noise that is present in mains powered equipment. My question is, could a unit like the 2302 be used in place of the typical 9V batteries in order to perform noise measurements (Sid, Svd)? I tend to think not...

Note: I haven't watched Dave's video the entire way through nor had a serious attempt at finding this information so my apologies if the answer is already evident.

[alm]

The noise is probably quite average for a linear power supply (pretty sure these are hybrid SMPS/linear, given their ability to provide more current at lower voltages). As far as I know Keithley does not specify the noise (not a very good sign), but the competing Agilent 663xx series was specified as 1 mV_RMS and 6 mV_p-p from 20 Hz - 20 MHz.

Pretty sure a few mV of noise will not have much effect on a cell phone given the amount of crap that goes on in there.

[Kleinstein]

The analog output stage with 2 FETs in series looks like it uses two input voltage each (e.g. +25,+15 V, 0 or -5 ?, -15 V).  I would expect it to work a little like a class H audio amplifier or some of the SMUs. So it is different from normal hybrid power supplies with a variable switched mode stage. They need the high power heat-sink for the 4 quadrant operation anyway.
It is slightly strange to have kind of 3 shunt values, but only 2 current ranges.

[alm]

The maximum source current is basically constant power (see attachment). Based on the specs, it looks suspiciously like a switching pre-regulator with a 6 V headroom for the linear regulator.

Not sure what that third shunt is used for. My guess is that it is for one of the extra current measurement modes, like the pulse current mode that can have a sample aperture down to 33 µs.

[EEVblog]

I like the use of high-res photos instead of live video for the detailed parts of the details.

I've only done that once before (but to even better effect) in the Siglent Spectrum Analyser teardown.


I did it this time because my Tagarno computer is down and I had to leave the lab early, so editing photos was the go.

People seem to like, so maybe I should do it more for PCB based teardowns like this ?

[alm]

I like it here because it provides a higher resolution, more favorable angles and a more stable picture (obviously ) than live camera work. The method lends itself well to a detailed analysis of a circuit. Obviously it is not suitable for every subject and should probably only be a limited part of the video lest the video becomes too static.

[EEVblog]

I like it here because it provides a higher resolution, more favorable angles and a more stable picture (obviously ) than live camera work. The method lends itself well to a detailed analysis of a circuit. Obviously it is not suitable for every subject and should probably only be a limited part of the video lest the video becomes too static.

Yep, agreed on all points.

[Fungus]

Seems like a LOT of components just to simulate a resistance.