EEVblog #591 - Agilent 54622D Retro Mixed Signal Osciloscope Review & Teardown

[EEVblog]

Dave checks out one of his favorite "old" oscilloscopes, the 2000 vintage Agilent 54622D mixed signal oscilloscope with the original megazoom ASIC technology and high resolution CRT display.
http://www.hit.bme.hu/~papay/edu/Lab/MegaZoom.pdf
http://cp.literature.agilent.com/litweb/pdf/5968-8152EN.pdf

[pickle9000]

So Dave what's the asking price?   

[EEVblog]

So Dave what's the asking price?   

No local price history on these in a long time. I sold my old work one about 5 years ago for $2400. Only paid the company $100 for it, suckers 

[EEVblog]

I was going to put it on ebay, but then realised that's probably not a good look after telling everyone to put it on their watchlist 

[EEVblog]

How many other 'scopes have you got to sell still?

This one, 3 Tek 3000's, and probably 15 or so TD220's in various states of distress.

[pickle9000]

I was going to put it on ebay, but then realised that's probably not a good look after telling everyone to put it on their watchlist 

Well you only sell the stuff in Australia so why not give some of the forum members the heads up and list it in a week or so. No matter what people will complain (about everything), I don't see it as an issue. How often do you get to see under the hood of an ebay item. From my perspective it's no different from buying a tshirt. If it helps keep the blogs coming then I say it's a good thing.

[Ketturi]

Speaking of scope screen resolution, when will those high resolution "retina" tablet displays come to scopes too? Yes it would need somewhat more processing speed, but waveform would like very crisp and and screen pixel layout would not interfere with it, no need for antialiasing.

Btw, would that used Agilent or similar one be better than these cheapest new DSOs for similar price? Yes older ones have less memory and bandwidth/sampling rate, but they looks to be much faster and better in the user interface side. What good is scope that promise the moon but are slow to use, display & UI is horrible and waveform looks crusty and jacked, no good for advanced signals. I ques good instruments age much better.
Edit: Well I obviously missed the end, where Dave said it would be worth

[KedasProbe]

Surprised to see such a relatively good analogue like scope view (on the AM signal)
To compare below my Hameg known for not being good in this area but a lot younger isn't even getting close:

[electronics man]

did i miss something, where were the ADC's

[marshallh]

Megazoom asics fabbed by Fujitsu

[Rutger]

They are listed on ebay for ~ $ 500 - $ 800. In that case they are not a good deal, you can get  a Rigol DS1074Z for $ 585 here in the US, that beats it by miles.

I like the vintage review, but when is Dave going to review the Rigol DS1000Z series, which in my option is a much better value for money scope.

[electronics man]

when is Dave going to review the Rigol DS1000Z series, which in my option is a much better value for money scope.

yeah Dave

[kevinpt]

I have seen them in good condition for $400 now and again. The key with the MSO versions is to find one that has the logic analyzer cable since they are often missing and the resellers gouge on them. The pods are shared with the other HP 40-pin LA cables so less of an issue there.

BTW I suspect that the gate array is handling the serial protocol trigger since that was new to these models compared to the 54645D.

Whatever you do Dave please don't get your hands on a LeCroy LT series since I'm waiting for the price to drop on those and a review would make them popular among the hobbyist crowd. That 9384 was a hunk of junk and all the other LeCroys are just as bad .

[Towger]

when is Dave going to review the Rigol DS1000Z series, which in my option is a much better value for money scope.

yeah Dave

I am also still waiting for the DS1000Z vs the the DS2000 series, there are a lot of different opinions over them. Even leaving out the obvious 2 vs 4 channels and cost.

[Hydrawerk]

Once I measured the waveform update rate of a similar Agilent scope.
EDIT: But it might be wrong for some reason (??). Probably an error in the external counter?

[Hydrawerk]

And then I made other videos.
XY mode

Asteroids game

Jumping letters

[EEVblog]

Once I measured the waveform update rate of a similar Agilent scope.

360Kwfms/sec, are you sure? I measured 500 or less.

[Hydrawerk]

Well, I am not too sure, and unfortunately I cannot measure it again, because that scope was built into a rack in the student's lab, so I cannot reach the Trig Out BNC any more.
But according to the datasheet this scope should be quite good at waveform update rate. There is  written

The MegaZoom deep memory is mapped into 32 levels of intensity on a display that has twice the horizontal resolution and update rates of up to 25 million vectors per second – so you see a more realistic representation of your signals.

This is on page 5 of the datasheet.

[Hydrawerk]

Once I measured the waveform update rate of a similar Agilent scope.

360Kwfms/sec, are you sure? I measured 500 or less.

You mean 500 wfms/sec or 500 000 wfms/sec?

[EEVblog]

You mean 500 wfms/sec or 500 000 wfms/sec?

500 wfms/sec, you can see the 500Hz figure on my scope in the video.

[EEVblog]

From my link before:
http://www.hit.bme.hu/~papay/edu/Lab/MegaZoom.pdf
Page 12:

Mega
Zoom
technology can deliver a maximum update rate of 7,800 waveforms per
second in the Infiniium 54830 Series oscilloscopes. In the 54640-series, the update
rate is maximum at 64,000 waveforms/second.

So that's the higher bandwidth 2GS/s 54640 models, not the 200MS/s models here.

[Hydrawerk]

:-( Uhh, I made some mistakes, just like this guy has made. http://www.ittsb.eu/GDS-2102A%20Wfms%20measurments.html
Maybe I can find another Agilent 546200 series scope and do the measurement again.
Well, thank you for clearing up this issue, in fact, I could hardly believe the 360 000 waveforms per second. It was quite high but still not unbelievable.

[adh]

The reason for specifing update rate in "vectors per second" and not waveforms/second is because LeCroy used to specify it in that way. For LeCroy scopes it made sense as they had real vector CRT's with update rate that depended on how complex screen content was.

[GBoos]

Hi Dave,

Agilent made their own optical Encoders, today you'll get them from AVAGO.
The Encoders look like HEDS-5700 without housing

[free_electron]

the encoders are not optical.
they are made by APEM if i remember correctly.

[mwilson]

I have an HP 54645D, the one Dave talks about early in the video as being the older model with the HP square-button styling. It's interesting to see this 54622D model (essentially the direct replacement of the 54645D, 2+16 channel 100MHz) -- I'd say it's a pretty significant upgrade over the previous model.

Most importantly is that the 54645D is a strictly monochrome display -- so no intensity-graded display of the waveform, the grid is the same intensity as the waveforms, etc. The 54645D also lacks serial triggering, the math functions are just A1+A2 or A1-A2 (no FFT, for example), and the serial/parallel ports were only available as an add-on module. Also no external trigger input.

Really neat to see the MegaZoom improvements as time went on from the 54645D to the 54622D and now all the way through the 2000X/3000X series. I have a 2001 Agilent catalog here (happened to find it in a box when I was cleaning recently) which lists the 54622D for US$4,495 at the time... not sure where inflation would put that at today. If they're going for less than $1,000 on eBay people are getting a pretty amazing scope for a great price. The 54645D shows its age relative to "modern" digital scopes, but it looks like the 54622D is much closer to the class of today's $4,500 scopes than its nearly 15-year age might lead folks to believe.

Thanks for the video, Dave!

[bg1cep]

just had a quick check of the 54624A update rate, and here are the relations to the timebase settings:

a: 5ns/div -> 360KHz update rate
b: 10ns/div -> 320K
c: 20ns/div -> 230K
d: 50ns/div -> 126.7K
e: 100ns/div -> 87.7K
f: 200ns/div -> 46.7K
g: 500ns/div -> about 2K

FYI

[mikeselectricstuff]

I have an HP 54645D, the one Dave talks about early in the video as being the older model with the HP square-button styling. It's interesting to see this 54622D model (essentially the direct replacement of the 54645D, 2+16 channel 100MHz) -- I'd say it's a pretty significant upgrade over the previous model.

Most importantly is that the 54645D is a strictly monochrome display -- so no intensity-graded display of the waveform, the grid is the same intensity as the waveforms, etc. The 54645D also lacks serial triggering, the math functions are just A1+A2 or A1-A2 (no FFT, for example), and the serial/parallel ports were only available as an add-on module. Also no external trigger input.

I think FFT comes when you add the serial /parallel port module. ISTR the serial is pretty poor as it only goes up to 19K2 due to it being inherited from an older model. If looking at this range, the 54622D is probably the earliest one to go for - seems a lot better than the 54645D, especially if you wnt to get data (even screenshots) out.

[kevinpt]

I think FFT comes when you add the serial /parallel port module. ISTR the serial is pretty poor as it only goes up to 19K2 due to it being inherited from an older model. If looking at this range, the 54622D is probably the earliest one to go for - seems a lot better than the 54645D, especially if you wnt to get data (even screenshots) out.

FFT is included in the GPIB module as well which would be the fastest way to extract data from the older 5464x lineup.
http://wiki.brown.edu/confluence/download/attachments/29211/hpstoragemod.pdf

[Hydrawerk]

Here is a screenshot from that presentation found by Dave Jones.

[Hydrawerk]

just had a quick check of the 54624A update rate, and here are the relations to the timebase settings:

a: 5ns/div -> 360KHz update rate
b: 10ns/div -> 320K
c: 20ns/div -> 230K
d: 50ns/div -> 126.7K
e: 100ns/div -> 87.7K
f: 200ns/div -> 46.7K
g: 500ns/div -> about 2K

FYI

I found another 54620 series scope at school and made the measurements again. The Trigger out frequency was about 330kHz at 5ns horizontal setting. I am kinda confused. If my measurements were wrong, then I really do not know why. Well, it is a mystery for me.

[Howardlong]

just had a quick check of the 54624A update rate, and here are the relations to the timebase settings:

a: 5ns/div -> 360KHz update rate
b: 10ns/div -> 320K
c: 20ns/div -> 230K
d: 50ns/div -> 126.7K
e: 100ns/div -> 87.7K
f: 200ns/div -> 46.7K
g: 500ns/div -> about 2K

FYI

I found another 54620 series scope at school and made the measurements again. The Trigger out frequency was about 330kHz at 5ns horizontal setting. I am kinda confused. If my measurements were wrong, then I really do not know why. Well, it is a mystery for me.

While I realise this is rather necroposting, I just did a brief test on a 54622D myself, and I get 362kwps with a 25MHz sine wave and 5ns/div, otherwise default settings.

However, I should note that this is not realtime sampling, it is operating in equivalent time sampling mode at this timebase setting by default.

Overriding to realtime sampling decreases the update rate to only 4kwps.

The automatic switch between equivalent time to realtime happens between 200ns/div and 500ns/div.

The maximum I seem to be able to achieve in realtime, with the 'right' signal and timebase is 4kwps with vectors, and about 12.3kwps with dots.

On a related note, this scope is super self-explanatory and simple to use, the menus are also simple and clear. The UI is also super quick. It could so easily be a benchmark for other scopes when it comes to ease of use. If I compare it to the Tek TDS460a I have here, admittedly a couple of years older, it seems that HP/Agilent and Tek took two completely different paths regarding UI design in those days with the advent of the DSO. HP's was a sleek, ergonomic design, whereas Tek seems to have struggled at that time, with a UI that is difficult to navigate and painfully slow to respond.

The 54622D I picked up appears to have been in storage unused for some time. It is an HP refurbished unit, and from the condition of the complete probe set that came with it, it appears to have rarely, if ever, been used since.  All the parts are there, and the channel identifying colour clips were not attached. When I first switched it on I had some trouble for a couple of hours with all the rotary encoders incorrectly registering, randomly going backwards and forwards as they were turned. After some vigorous turning backwards and forwards and a bit of time, a couple of hours later they were fine. It was a cold day, and I strongly suspect that judging from the screen condensation, the encoders had become temporarily contaminated too.

But yes, I can see why these puppies were so popular. These days, I think that the 54624A 4 channel analogue-only may be a better bet: I seem to use LAs only rarely these days, and before there were LAs, I used a scope anyway, even for 16 bit processors (that was indeed a lesson in patience debugging your software by probing the address and data lines with a scope!) The I2C triggers fine on the analogue channels, but regretfully there is no I2C decode, however for us old farts who've been doing I2C since the 80s it's not such a big deal, but the deep memory together with I2C triggering is very nice.

So,if you happen to come across one of these beauties at a reasonable price, you might just enjoy using it.

[Hydrawerk]

While I realise this is rather necroposting, I just did a brief test on a 54622D myself, and I get 362kwps with a 25MHz sine wave and 5ns/div, otherwise default settings.

However, I should note that this is not realtime sampling, it is operating in equivalent time sampling mode at this timebase setting by default.

Overriding to realtime sampling decreases the update rate to only 4kwps.

The automatic switch between equivalent time to realtime happens between 200ns/div and 500ns/div.

The maximum I seem to be able to achieve in realtime, with the 'right' signal and timebase is 4kwps with vectors, and about 12.3kwps with dots.

Thank you for the information.

[kd4ttc]

I just picked one up on eBay for $595+ $35 shipping. Included manuals, digital harness and analog probes. Beautiful scope.  Very intuitive operation. Manuals online at Keysight. Others available on ebay same price range, though I agree with other comment, be sure you get one with the digital probes. Of course, upgrade was huge over my old HP180.

[Someone]

just had a quick check of the 54624A update rate, and here are the relations to the timebase settings:

a: 5ns/div -> 360KHz update rate
b: 10ns/div -> 320K
c: 20ns/div -> 230K
d: 50ns/div -> 126.7K
e: 100ns/div -> 87.7K
f: 200ns/div -> 46.7K
g: 500ns/div -> about 2K

FYI

I found another 54620 series scope at school and made the measurements again. The Trigger out frequency was about 330kHz at 5ns horizontal setting. I am kinda confused. If my measurements were wrong, then I really do not know why. Well, it is a mystery for me.

Dont worry, the measurements in Dave's video are wrong as there is a substantial pre trigger delay limiting the update rate.

[Howardlong]

Yesterday, a 54642D arrived here, this is essentially the same as a 54622D but with 500MHz bandwidth and 2GSa/s.

I noticed that in the triggering menu of the 54642D there are a lot more options for serial triggering, including SPI, LIN, USB as well as I2C. This 54642D also has the optional CAN trigger. My 54622D only has I2C.

When looking at the Agilent catalogs, it would appear that early units only offered I2C, then around 2002, the additional serial triggering modes were offered. The 54622D software/ROM revisions are A.01.00 and A.01.10 dated 2000, whereas the 54642D is A.02.20 and A.02.00 and appears to be from 2002. The paper manual that came with the 54622D (an Agilent refurb, Feb 2001) also only mentions I2C whereas the later online manual for these scopes discusses all of the serial options.

I am wondering if the earlier 2000 units are firmware upgradeable, or if there were some hardware changes too?

[Howardlong]

In answer to my own question, the firmware is available for both the 54622D and 54642D from the Agilent/Keysight website, not sure why I didn't find it before. Go to the product page, then select the "Visit Technical Support" link, then select the Drivers, Firmware and Software tab. The firmware is in that list.

In addition it looks like for full CAN triggering you need a trigger module. It will trigger on SOF without, but that appears to be it for CAN.

[kd4ttc]

There is an expansion port on the back of this scope (Agilent 54622D).  Does anyone know the specs on the connector and pin outs?

[kd4ttc]

I bought a 54622D a few months ago off ebay in beautiful condition with digital probes and manuals for about $650.  It really is a lovely scope. I'm just using it for hobby digital electronics use, but my son is graduating from Pirdue this year as an EE and he was home this weekend and had a little project to do. He was measuring circuit chatacteristics of a design for a stepper motor control.

Wow, so intuitive of an interface. He's used scopes at Purdue and at S&C Electric where he did his Co-op, so he knows a scope. Buy the ease he had working the controls without having read the manual was remarkable.  I'd read the manual so knew to point out the measurement functions, but with that one button push he was up and running with that, too.

This scope has a 16 channels of digital inputs as well.  Before I purchased the scope I searched around and found the digital cable for this unit is about $200-250. If you're buying look for a unit with the digital input harness. I imagine that a unit with the cables would probably have been better cared for than one where the cable was lost.  The digital inputs proved useful for an additional trigger input in the current project.

I appreciate Dave having reviewed this scope.  His review has highlighted a very capable instrument.

[free_electron]

These machines are indeed the bees knees.
I got several very expensive scopes (7104 mso, 54832d's). But whenever i need something debugged quickly : grab the 54622 or 54645.
They start up fast, react swiftly to button operation and they get the job done

[Howardlong]

Thanks for doing the pull apart Dave .

Got one of these on the way and as usual never looked about before i clicked the button , it looked good enough and now it looks even better too .

Cheers

The biggest problem with these scopes is that both the UI and responsiveness are so good you just want to buy another, newer one. And then another. Unlike Microsoft, HP/Agilent/Keysight realised that once you have such a great UI you'd be nuts to want to re-invent it, and that UI is still the same today albeit with a bit of colour, so going from a 54622D to an x3000 is simplicity itself.

[Alexandr]

Does anyone have any power supply circuit Agilent 54622d? I want to change burned-out elements?

[SilverSolder]

[...] regretfully there is no I2C decode

I also have a well loved 54622D...  an excellent "old school product" that actually feels quite modern to work with as you say.

If you are desperate to decode I2C or other protocols:   In some cases you can export the sampled waveforms via floppy and decode them on your PC with a small program (that you write) -  this technique works for other serial (and parallel) protocols too, including non standard protocols - but does require some quality time in front of the computer of course.

[Howardlong]

[...] regretfully there is no I2C decode

I also have a well loved 54622D...  an excellent "old school product" that actually feels quite modern to work with as you say.

If you are desperate to decode I2C or other protocols:   In some cases you can export the sampled waveforms via floppy and decode them on your PC with a small program (that you write) -  this technique works for other serial (and parallel) protocols too, including non standard protocols - but does require some quality time in front of the computer of course.

For many of us, we spent many years manually decoding anyway, so although serial decode is a great feature to have, it's not essential. In fact having a working knowledge of how your serial busses should look, especially regarding timing, is a very useful skill to have anyway, without relying on the decode function (which may not even be set up correctly anyway).

[nctnico]

You'll also trade your car for a horse & carriage? 

[Howardlong]

You'll also trade your car for a horse & carriage? 

Indeed, it makes you wonder how we ever managed!

[SilverSolder]

If you work in a commercial environment where you need to turn work around quickly, a modern scope with exactly the right level of capability is a rational choice. 

It is also rational for a hobbyist if you can afford it (no worse than blowing $2K on a fancy camera lens or something, right?).

Keeping older equipment alive is also rational if it does the job you need it to do   It can also be irrational of course, but fun...   like buying a used red Porsche and getting a wig!   

[Robomeds]

just had a quick check of the 54624A update rate, and here are the relations to the timebase settings:

a: 5ns/div -> 360KHz update rate
b: 10ns/div -> 320K
c: 20ns/div -> 230K
d: 50ns/div -> 126.7K
e: 100ns/div -> 87.7K
f: 200ns/div -> 46.7K
g: 500ns/div -> about 2K

FYI

I found another 54620 series scope at school and made the measurements again. The Trigger out frequency was about 330kHz at 5ns horizontal setting. I am kinda confused. If my measurements were wrong, then I really do not know why. Well, it is a mystery for me.

While I realise this is rather necroposting, I just did a brief test on a 54622D myself, and I get 362kwps with a 25MHz sine wave and 5ns/div, otherwise default settings.

However, I should note that this is not realtime sampling, it is operating in equivalent time sampling mode at this timebase setting by default.

Overriding to realtime sampling decreases the update rate to only 4kwps.

The automatic switch between equivalent time to realtime happens between 200ns/div and 500ns/div.

The maximum I seem to be able to achieve in realtime, with the 'right' signal and timebase is 4kwps with vectors, and about 12.3kwps with dots.

On a related note, this scope is super self-explanatory and simple to use, the menus are also simple and clear. The UI is also super quick. It could so easily be a benchmark for other scopes when it comes to ease of use. If I compare it to the Tek TDS460a I have here, admittedly a couple of years older, it seems that HP/Agilent and Tek took two completely different paths regarding UI design in those days with the advent of the DSO. HP's was a sleek, ergonomic design, whereas Tek seems to have struggled at that time, with a UI that is difficult to navigate and painfully slow to respond.

The 54622D I picked up appears to have been in storage unused for some time. It is an HP refurbished unit, and from the condition of the complete probe set that came with it, it appears to have rarely, if ever, been used since.  All the parts are there, and the channel identifying colour clips were not attached. When I first switched it on I had some trouble for a couple of hours with all the rotary encoders incorrectly registering, randomly going backwards and forwards as they were turned. After some vigorous turning backwards and forwards and a bit of time, a couple of hours later they were fine. It was a cold day, and I strongly suspect that judging from the screen condensation, the encoders had become temporarily contaminated too.

But yes, I can see why these puppies were so popular. These days, I think that the 54624A 4 channel analogue-only may be a better bet: I seem to use LAs only rarely these days, and before there were LAs, I used a scope anyway, even for 16 bit processors (that was indeed a lesson in patience debugging your software by probing the address and data lines with a scope!) The I2C triggers fine on the analogue channels, but regretfully there is no I2C decode, however for us old farts who've been doing I2C since the 80s it's not such a big deal, but the deep memory together with I2C triggering is very nice.

So,if you happen to come across one of these beauties at a reasonable price, you might just enjoy using it.

I know this is a old thread but...

I picked up one of the 54621A scopes recently.  I used the older 54600 scopes (60Mhz, played Tetris) years back and about 10 years back helped the lab I was in pick up a MSO6000 series scope.  So even though I used the two families of scopes around this one, I hadn't actually used one of these.  A while back I picked up a Siglent based scope for personal use.  I was never really that happy with it. The UI was fine but the scope was just seriously laggy.  I actually preferred using the old Tek TDS 220 based scopes much of the time because they had less lag.  The Agilent is just great in that regard.  Mine boots in 5 seconds. 

Anyway, I was curious about the waveform update rate Dave reported.  Since I don't have a nice signal generator I used an analog NTSC video signal (composite video) as a source.  I showed an 87khz trigger rate (vectors, not realtime)  Using real time reduced things to about 3.5khz with vectors (didn't think to test with vectors off). 

Great low buck hobby scope!
 

[Howardlong]

I picked up one of the 54621A scopes recently.  I used the older 54600 scopes (60Mhz, played Tetris) years back and about 10 years back helped the lab I was in pick up a MSO6000 series scope.  So even though I used the two families of scopes around this one, I hadn't actually used one of these.  A while back I picked up a Siglent based scope for personal use.  I was never really that happy with it. The UI was fine but the scope was just seriously laggy.  I actually preferred using the old Tek TDS 220 based scopes much of the time because they had less lag.  The Agilent is just great in that regard.  Mine boots in 5 seconds. 

Anyway, I was curious about the waveform update rate Dave reported.  Since I don't have a nice signal generator I used an analog NTSC video signal (composite video) as a source.  I showed an 87khz trigger rate (vectors, not realtime)  Using real time reduced things to about 3.5khz with vectors (didn't think to test with vectors off). 

Great low buck hobby scope!
 

These scopes really are such a joy to use, the UI is just so straightforward and responsive. It's a canonical lesson for UI designers everywhere if ever there was one.

It's been a while since I did any waveform update rate tests on this scope, but you'll need to have the right trigger repetition rate to maximise the update rate.

[SilverSolder]

As a first look:    If the green lights/buttons do their "boot up song and dance" when you start the scope, the CPU is probably working fine and it could well be a CRT section problem as you suggest.

If everything remains stone cold dead when you press the power button, perhaps it is a power supply issue or other problem.