How does Agilent do 80 GSa/s?

[IF]

Hi,

I cannot wrap mind around the new line of Agilent scopes, the 90000  X-Series. How did they manage to get 80 GSa/s of real time sampling?
Is this a similar trick with multiple AD converters, Rigol used in their scopes?

Cheers!
z.

[alm]

High end real-time scopes use both interleaving (either in time or frequency domain) and really fast ADC's with a proprietary semiconductor process (no overclocking standard AD parts at these speeds). Some of the past high-end Agilent scopes had 80 interleavers per channel, not sure about the current ones.

[mikeselectricstuff]

They use  a custom Indium Phospide front-end IC : Datasheet with cool pic of chip
The price tag on this scope is well into 6 figures - even the probes are $20K each!
I notice the various models only differ by bandwidth - I wonder if the ASIC includes a bandwidth limiter like the Rigol scope

[safarir]

They use  a custom Indium Phospide front-end IC : Datasheet with cool pic of chip
The price tag on this scope is well into 6 figures - even the probes are $20K each!
I notice the various models only differ by bandwidth - I wonder if the ASIC includes a bandwidth limiter like the Rigol scope

Probably, in the datasheet, they say that the bandwitch is fully upgradable...
"Buy the performance
you need today
knowing you have the
headroom you need
for tomorrow with
bandwidth upgradability
to 32 GHz"

[TheWelly888]

At those prices, they must be manufactured to order and there's no way our Dave will get to "Take it Aparrt!" in his garage!!!

[Polossatik]

*drool*

[jahonen]

My experience on the higher end oscilloscopes has been that although the specs and features are nice, they are quite cumbersome for everyday use. Alone that fact that they usually have some kind of Windows built in, and that takes the time to boot up. Very annoying when you want to just make couple of simple checks. I already tend to get frustrated when my R&S boots up the Windows when I want to do some spectrum measurements Not to even speak about when you have the scope connected to the LAN and then Windows wants to do the reboot after installing some updates when you are in the middle of measuring something. You even can't connect ordinary probes easily into those. If you look the connectors on the 90000 series, no way you could use the BNC directly anymore. Adapter required, probably at extra cost.

That means those ordinary measurements where you don't need the extreme GHz bandwidth, I much more prefer the mid-range stuff, like Agilent 5000 or 6000 series (no Windows!).

Of course, if one does high-speed stuff, this is something which one needs to validate and characterize the design. But usually that is way out of average hobbyist stuff.

Regards,
Janne

[ArtemisGoldfish]

My experience on the higher end oscilloscopes has been that although the specs and features are nice, they are quite cumbersome for everyday use. Alone that fact that they usually have some kind of Windows built in, and that takes the time to boot up. Very annoying when you want to just make couple of simple checks. I already tend to get frustrated when my R&S boots up the Windows when I want to do some spectrum measurements Not to even speak about when you have the scope connected to the LAN and then Windows wants to do the reboot after installing some updates when you are in the middle of measuring something. You even can't connect ordinary probes easily into those. If you look the connectors on the 90000 series, no way you could use the BNC directly anymore. Adapter required, probably at extra cost.

That means those ordinary measurements where you don't need the extreme GHz bandwidth, I much more prefer the mid-range stuff, like Agilent 5000 or 6000 series (no Windows!).

Of course, if one does high-speed stuff, this is something which one needs to validate and characterize the design. But usually that is way out of average hobbyist stuff.

Regards,
Janne

Ugh, I *hate* Windows based test equipment. I've had occasion to use a Sencore RF analyzer that runs Windows NT 4.0 on an AMD K6, and it's just so annoying. It takes 5 minutes to boot up, whereas an actual RF analyzer not based on windows made by the same company takes about...20 seconds to boot.

[alm]

My experience on the higher end oscilloscopes has been that although the specs and features are nice, they are quite cumbersome for everyday use. Alone that fact that they usually have some kind of Windows built in, and that takes the time to boot up. Very annoying when you want to just make couple of simple checks. I already tend to get frustrated when my R&S boots up the Windows when I want to do some spectrum measurements Not to even speak about when you have the scope connected to the LAN and then Windows wants to do the reboot after installing some updates when you are in the middle of measuring something.

I'm not very happy about a heavy, fragile OS on expensive test equipment either, but the customers want to run Excel *shiver* and Matlab on them.

You even can't connect ordinary probes easily into those. If you look the connectors on the 90000 series, no way you could use the BNC directly anymore. Adapter required, probably at extra cost.

Not very surprising, ever looked at the performance of BNC at 20GHz? You need a higher performance connector at those speeds, plus you need all the power and communication pins for active probes. I'm sure the manufacturers don't mind that you are forced to buy their probes. The low-impedance BNC adapters are only a few hundred dollars or so I believe, that's peanuts compared to the price of the scope.

[jahonen]

Not very surprising, ever looked at the performance of BNC at 20GHz? You need a higher performance connector at those speeds, plus you need all the power and communication pins for active probes. I'm sure the manufacturers don't mind that you are forced to buy their probes. The low-impedance BNC adapters are only a few hundred dollars or so I believe, that's peanuts compared to the price of the scope.

I did not mean that 20 GHz scope should have BNC connectors, which would be absolutely absurd thing to do. One Agilent guy who lectured about their RF products, like spectrum and network analyzers said that BNC is not even a connector by his standards, maybe good for 100 kHz or so

But meant instead that those scopes are not so handy for everyday measurements where you can manage with fraction of the bandwidth and often need higher voltages than these things can directly handle. That is true that passive probe adapter costs next to nothing when compared to the scope but it still makes things clumsy in practice, compared to simpler scope.

Regards,
Janne

[alm]

I did not mean that 20 GHz scope should have BNC connectors, which would be absolutely absurd thing to do. One Agilent guy who lectured about their RF products, like spectrum and network analyzers said that BNC is not even a connector by his standards, maybe good for 100 kHz or so

Darn, not even good enough for audio (in the minds of audiophiles)

But meant instead that those scopes are not so handy for everyday measurements where you can manage with fraction of the bandwidth and often need higher voltages than these things can directly handle. That is true that passive probe adapter costs next to nothing when compared to the scope but it still makes things clumsy in practice, compared to simpler scope.

Agreed. Passive (hi-Z) probes only go up to 500MHz, so if you use one, you probably don't need many GHz of bandwidth. And entry/mid level scopes are indeed much more practical for everyday use. Plus they take longer to get obsolete, while the high end products are typically obsolete by the time the next serial bus comes along.