Tektronix Mixed Domain MDO4000 Review

[FreeThinker]

Yes but $20k+ for a scope with flakey Usb?. Either they are faulty and need sorting or they are badly specced and need sorting But it needs a good coat of looking at before they batch any more! Yes?

[IanB]

my guess is the eval board came after they'd specced the USB current limit

Are there any USB ports these days that can't supply 500 mA? Especially on devices with ample on-board power?

I really don't think there is any excuse for restricted current.  It would seem like a major goof-up.

As Dave observes in the review, it seems like the marketing department have grafted on this spectrum analyzer feature for feature points in the sales literature, but they have not allowed the engineers to improve the basic design of the oscilloscope in a way that would make it appealing to the engineering types who might want to use it as a practical working tool. It's a cool idea badly executed.

[Zad]

Thanks for the photos Dave!

[EEVblog]

Awhile back user allanw posted some pictures of a DPO4034, it's interesting to see how that's different from this one

https://www.eevblog.com/forum/index.php?topic=1773.0
http://picasaweb.google.com/113939095023945686646/TektronixDPO4034?authkey=Gv1sRgCLDw-ZCnsqCdkgE

In the DPO4034, they're using a lot of off the shelf components, a Stratix FPGA and AMCC PPC440EP applications processor. They probably use a similar processor in the MDO.

The most interesting thing is the ATI Mobility Radeon GPU. I bet the big chip in the MDO is also an ATI GPU. They're probably using the parallel processing capabilities to build the graded intensity waveform display and do math/waveform processing, but I suspect the performance isn't as good as Agilent's ASICs, hence why this scope's update rate isn't as good. Also of note is that the clock speed of the PPC440 is either the same or faster than the CPU in the Agilent X series, yet the Agilent is much more responsive. Must be either careless programming on Tek's part, or limitations due to using the GPU for waveform processing. Heck, even the old Agilent 5462x series from the mid 90s were as responsive as the new Agilent stuff, using a 14MHz processor!

I used to have a 54621D and it was a great scope, and was certainly as responsive as anything going today.

Dave.

[EEVblog]

my guess is the eval board came after they'd specced the USB current limit

Are there any USB ports these days that can't supply 500 mA? Especially on devices with ample on-board power?

I really don't think there is any excuse for restricted current.  It would seem like a major goof-up.

I agree, it's a goof-up, and having to plug in both to make it work is a kludge.

Dave.

[Thomas]

I think the two USB connectors are there for a reason: it needs more power than can be provided from one USB port.
Seen it on some external HDD's and some instruments.

[ejeffrey]

Are there any USB ports these days that can't supply 500 mA? Especially on devices with ample on-board power?

I really don't think there is any excuse for restricted current.  It would seem like a major goof-up.

The only common case is bus powered USB hubs (including those built in to peripherals like keyboards).  Since USB requires power negotiation, it the failure is probably not actually inability to provide sufficient current, but that it is not negotiating correctly.  My WAG would be that their USB controller didn't support as many ports as they wanted so they threw an off-the-shelf USB hub chip in but forgot to to wire the pin (or whatever) to tell the hub controller there is external power available.

However, this doesn't really explain the problem with the USB flash drives which I thought were usually low power devices.  Maybe that is a separate issue with the filesystem driver.

[EEVblog]

I think the two USB connectors are there for a reason: it needs more power than can be provided from one USB port.
Seen it on some external HDD's and some instruments.

I haven't measured it, but it works fine on just one USB port on the Agilent scope, and every other USB port I try that can supply the usual 500mA (the "negotiation" requirement is a misinterpreted myth).
Tek put in wimpy USB power, and my guess is they realised it after the fact, so provided the split USB cable.

Dave.

[jahonen]

I think the two USB connectors are there for a reason: it needs more power than can be provided from one USB port.
Seen it on some external HDD's and some instruments.

I haven't measured it, but it works fine on just one USB port on the Agilent scope, and every other USB port I try that can supply the usual 500mA (the "negotiation" requirement is a misinterpreted myth).
Tek put in wimpy USB power, and my guess is they realised it after the fact, so provided the split USB cable.

Dave.

What myth, the USB 2.0 specification is quite clear on that point (unit load = 100 mA).

I personally hate those ignorant "just ignore the spec, it'll work fine" guys. I know the negotiation is a PITA on simple devices but it is the only way to guarantee interoperability. While most hobbyists do not care but I'd expect that at least Tek guys wouldn't be so ignorant.

7.2.1 Classes of Devices

The power source and sink requirements of different device classes can be simplified with the introduction of the
concept of a unit load. A unit load is defined to be 100 mA. The number of unit loads a device can draw is an
absolute maximum, not an average over time. A device may be either low-power at one unit load or highpower,
consuming up to five unit loads. All devices default to low-power. The transition to high-power is under
software control. It is the responsibility of software to ensure adequate power is available before allowing
devices to consume high-power.

and

7.2.1.4 High-power Bus-powered Functions

A function is defined as being high-power if, when fully powered, it draws over one but no more than five unit
loads from the USB cable. A high-power function requires staged switching of power. It must first come up in
a reduced power state of less than one unit load. At bus enumeration time, its total power requirements are
obtained and compared against the available power budget. If sufficient power exists, the remainder of the
function may be powered on. A typical high-power function is shown in Figure 7-45. The function’s electronics
have been partitioned into two sections. The function controller contains the minimum amount of circuitry
necessary to permit enumeration and power budgeting. The remainder of the function resides in the function
block. High-power functions must be capable of operating in their low-power (one unit load) mode with an
input voltage as low as 4.40 V, so that it may be detected and enumerated even when plugged into a buspowered
hub. They must also be capable of operating at full power (up to five unit loads) with a VBUS voltage
of 4.75 V, measured at the upstream plug end of the cable.

I think the present situation is due that most host devices have no proper power management. I know that my friend once smoked usb cable of his camera when the camera end connector shorted out. So much for even 500 mA current limit. Probably the motherboard did not contain any kind of current limiting, thus passing through all what was available on the PC PSU.

Regards,
Janne

[Bored@Work]

(the "negotiation" requirement is a misinterpreted myth).

Then explain what the USB standard, Revision 2.0, April 27, 2000 means in section 7.2, especially section 7.2.1. and 7.2.1.4. Lets just take a look at parts of them:

7.2.1 Classes of Devices

The power source and sink requirements of different device classes can be simplified with the introduction of the concept of a unit load. A unit load is defined to be 100 mA. The number of unit loads a device can draw is an absolute maximum, not an average over time. A device may be either low-power at one unit load or highpower, consuming up to five unit loads. All devices default to low-power. The transition to high-power is under software control. It is the responsibility of software to ensure adequate power is available before allowing devices to consume high-power.
...

7.2.1.4 High-power Bus-powered Functions

A function is defined as being high-power if, when fully powered, it draws over one but no more than five unit loads from the USB cable. A high-power function requires staged switching of power. It must first come up in a reduced power state of less than one unit load. At bus enumeration time, its total power requirements are obtained and compared against the available power budget. If sufficient power exists, the remainder of the function may be powered on. A typical high-power function is shown in Figure 7-45. The function’s electronics
have been partitioned into two sections. The function controller contains the minimum mount of circuitry necessary to permit enumeration and power budgeting. The remainder of the function resides in the function block. High-power functions must be capable of operating in their low-power (one unit load) mode with an input voltage as low as 4.40 V, so that it may be detected and enumerated even when plugged into a buspowered hub. They must also be capable of operating at full power (up to five unit loads) with a VBUS voltage of 4.75 V, measured at the upstream plug end of the cable.

...

The requirement is not a myth, it is a reality. Completely in the open for everyone who cares to read the USB standard.

Which is the point "cares to read the USB standard". The problem are developers running around, designing USB devices and giving a toss about the USB standard, declaring requirements of it a myth, while they haven't read a single line of the standard.

Most device designers give a toss, and use as much power they dammed like, and most host designers have bend over and just supply whatever power the host wants until the USB host chip goes up in flames. But that doesn't change the requirements in the standard. The requirements are real.

And the requirements can be summarized as 0 mA on bus suspension and 100 mA max. on device power up. Then the device specifies its requirement in the bMaxPower field in the configuration descriptor, and gets the required power granted with a Set config  or denied.

But yes, it is more convenient running around and wanking over a "weak" USB host port. The truth is, you know nothing about the port.  You would need a logic analyses with USB decoder and current measurement to figure out what is really going on. Does the device board properly enumerate? Does it properly not use more than 100 mA at start? Does it properly specify its required power in bMaxPower in its device configuration descriptor? Does the host grant that power?

Without having done those measurements you know nothing, especially not if it is the oscilloscope's or the board's fault. And don't give me that "I have 20 years of experience", or "everyone does it". Read the standard, measure, compare the results against the standard and then we are talking. Not that childish "they designed a weak port". Maybe they did. But you don't know.

[Bored@Work]

What myth, the USB 2.0 specification is quite clear on that point (unit load = 100 mA).

Good to see I am not the only person in the world who has read the USB standard.

[ejeffrey]

I haven't measured it, but it works fine on just one USB port on the Agilent scope, and every other USB port I try that can supply the usual 500mA (the "negotiation" requirement is a misinterpreted myth).
Tek put in wimpy USB power, and my guess is they realised it after the fact, so provided the split USB cable.

It is not a myth, although it is often misinterpreted.  It is true that USB ports almost never have programmable current limiting.  If you attach a 10 ohm resistor across the power pins of a standard USB port, you are going to get 500 milliamps of current but be in violation of the spec.  To be in compliance you must request high power mode.  If you do not, there is every chance that not only will your device fail to work when plugged into a bus-powered hub, but you take down other devices plugged into the same hub.  There is really no excuse to not do proper USB power negotiation.

If the demo board is operating correctly it will power up the microcontroller on insertion, then only power on the rest of the circuitry after requesting high-power mode from the host.  This is why I think it is more likely solution is not that the 5V rail is underpowered, but the host is failing to negotiate high power devices.

On the other hand, it is certainly possible that they put an underpowered regulator on the USB5.  It is even possibly they advertise high power capability but don't actually have a beefy enough supply to support it.  I kind of doubt this it is easier to screw up the protocol than the power supply.

[EEVblog]

It is not a myth, although it is often misinterpreted.  It is true that USB ports almost never have programmable current limiting.  If you attach a 10 ohm resistor across the power pins of a standard USB port, you are going to get 500 milliamps of current but be in violation of the spec.

And that is exactly what I am talking about.
I'm not talking about the proper design of USB devices, I'm talking ONLY about the USB ports ability to supply 500mA.
The common "myth" I referred to is that a USB port will ONLY supply 100mA max UNTIL an intelligent device requests more.
AFAIK, that is NOT the case. A USB port will be able to supply the 500mA, and they do, without negotiation.

Talk about proper USB device design is an entirely different issue, and you won't get any argument from me.

Dave.

[ejeffrey]

Maybe you could hook your constant current load up to the USB power pins and see how much you can draw before it starts to drop out?  That would answer the question whether it was a protocol issue or a power supply issue.

[EEVblog]

Maybe you could hook your constant current load up to the USB power pins and see how much you can draw before it starts to drop out?  That would answer the question whether it was a protocol issue or a power supply issue.

Ok, I went and did some measurements, mystery solved.

The demo board draws around 850mA, so clearly is not designed to be operated from a single USB port, hence the dual plug supplied. The red "power only" plug is included to get the extra power from two paralleled ports, and not to simply power the device without any comms as I had assumed.

The Tek USB port on the front can supply up to around 800mA before the protection circuit kicks in (probably lower if left longer), hence the hiccuping observed. All the other USB ports I used were obviously able to supply that 850mA without the protection circuit kicking in, hence the confusion and my assumption that the board was taking 500mA max.

So the Tek ports aren't "weak" as I lead to suspect.

Dave.

[EEVblog]

For those interested, Iiro Sundberg emailed me an attempt at a system description of the RF section, with labels overlayed on an image of the board:

I am an RF engineer myself but I will not guarantee that I have got everything right. I have assumed that after the Panasonic RF Relay (ARJ224H or something) there is actually a mixer since it seems that there is a oscillator of some sort and an amplifier (311 if I saw it correctly). After this there is three pieces of a pi-attenuators which can be passed by switches (09V-marked components) and eventually some amplification is available if needed (can be passed also). After this point it gets exciting. There is actually three different channels which I assume are for different frequencies. Just a hunch but the upper most could be lowest in frequency and the high filter or low filter channels are then down-converted to a intermediate frequency close to the upper most channel (or ADC or what ever is the detection scheme) depending the switch settings. Next there is the logaritmic amplifier which is probably used to take a sample of the power level and then the input attenuators are adjusted so that the receiver does not compress. Again just a guess but in that ferrite thingy is a coaxial cable running through and it somehow divides the signal into two channels again. After that there is more amplification and some filtering. Then there is again a Variable Controlled Oscillator but I am little puzzled what is going on there. Maybe the signals are routed to ADC through VIA and an other layer. The Printed Wiring Board material is probably something like Rogers RO4000 series.

[ejeffrey]

For those interested, Iiro Sundberg emailed me an attempt at a system description of the RF section, with labels overlayed on an image of the board:

Nice!

[Chasm]

Agilent were infinitely smarter than Tek though, they let me keep it so it took pride of place on my shelf, and now gets in almost every video for years to come, offering untold ROI.
Tek will pay millions in full page traditional media ads that are gone in the flick of a page
Gossen were the first to see the value right back the start of the blog, giving me those meters early on in the hope they get in my videos.

I suppose that Tek would rather give the scope to someone who actually does RF work.
So, are there any ham related video podcasts out there? If not it may be high time to start one. Maybe yet another project for Chris?

[metalphreak]

Every modern PC motherboard i've looked at, has the USB power pins connected directly to the systems 5V power rail, with protection in the form of a 1A surface mount PTC fuse. Its almost always 1 PTC per 2 USB ports, so 500mA each. It's really only meant for short circuit protection, but this means you can easily draw more than 500mA from a PC's USB port. Many older PCs had no current limiting protection, and a dead short meant a fried motherboard. It's up to the device to conform to specs and tell the PC what its drawing.

[alexwhittemore]

That's a rough statement, 'most modern boards.' I know from experience that my MBP (notorious for lower-power USB ports, more like 550 mA than 850) gets VERY mad when you overcurrent it. The OS complains, that USB host shuts down, won't come on till proper data negotiation...

[Hydrawerk]

Well, is it true that Tektronix hasn't introduced any new scope since this in 2011??
Except for the funny "new" TBS1000 series... 

[EEVblog]

Well, is it true that Tektronix hasn't introduced any new scope since this in 2011??
Except for the funny "new" TBS1000 series... 

Yes, it's true. And nobody expects them too any time soon.
The "Dananher System" probably has the R&D team (what's left of them) by the short and curlies.

[w2aew]

Well, is it true that Tektronix hasn't introduced any new scope since this in 2011??
Except for the funny "new" TBS1000 series... 

Yes, it's true. And nobody expects them too any time soon.
The "Dananher System" probably has the R&D team (what's left of them) by the short and curlies.

Not really true. A few additions to the MDO4000 series were introduced several months ago, a new DSA8300 sampling platform was introduced last year, and there were two ultra-performance scopes introduced recently at 25 and 33GHz bandwidth (100GS/s).  Nothing really new on the low end, but I'm sure that will change in the future.

In addition to that, there have been two new RSA5100A series real-time spectrum analyzers (15 and 26.5GHz), and a pair of new Arbitrary Waveform Generators (2ch @ 25GS/s and 1ch @ 50GS/s), all introduced this year.

[EEVblog]

Not really true. A few additions to the MDO4000 series were introduced several months ago, a new DSA8300 sampling platform was introduced last year, and there were two ultra-performance scopes introduced recently at 25 and 33GHz bandwidth (100GS/s).  Nothing really new on the low end, but I'm sure that will change in the future.

A new low end Tek scope, the Lochness monster of the T&M world! 

[Wuerstchenhund]

and there were two ultra-performance scopes introduced recently at 25 and 33GHz bandwidth (100GS/s).

Yes, but I'm not sure that will help Tek much. The DSO/DSA700000 competes with Agilent's DSOX9000Q and LeCroy's LabMaster 9Zi/10Zi, which offer higher sampling rates (120/160GSa/s, and the LeCroy even does that on all channels), and all are available with higher bandwidths than Tek offers as well. Considering that the Tek isn't even cheaper than Agilent and LeCroy, there is probably very little reason to go with Tek nowadays.

Dave is completely right with his comments re. Danaher. Their 'Danaher Business System' (DBS) is focused on shareholder value and micromanagement, and offers very little room for creativity or taking risks (which is what most new technologies are). The result is a Tek that is constantly behind the curve and outperformed by its competition. Their offerings are depressing, and it seems their sales staff is aware of that as they have become much more active recently (I never got so many calls from our Tek representatives trying to get some business).

I grew up with a lab full of Tek, but nowadays I would not even consider them anymore, neither for a low end scope and even less for a high end scope.