Analog multipliers and watt measurement

a

uff

Woulds you care to expand that sentence? A/D conversion is the same process in every application. The importance of the errors involved may differ from application to application, but reducing those errors is always a good idea.

r 50K

d the

code,

n r

o-

-

It's possible that the GaAs did too, by ending u with a lower dislocation density but I never saw any doci=3Dumented evidence supporting that idea.

a e

You flatter yourself - not for the first time. It happens often enough to irritate you into posting petulant nonsense like " How come you always know how to do everything better than other people?".

d

y,

A familiar kind of idiocy. I was wrong to imagine that the commodity market aspect of the situation was paramount.

So - none of the right kind of in-house technical talent, like I said

As opposed to the money you got for developing the product that they did sell.

-- Bill Sloman, Nijmegen

The other URL that I posted

includes a reference to a paper on dithering in audio A/D conversion that adresses exactly that kind of point, albeit in passing

R. M. Gray and T. G. Stockman, IEEE Trans. Inf. Theory IT-39, 805 (1993).

Conversion errors can be inconvenient for all sorts of reasons.

-- Bill Sloman, Nijmegen

All good stuff, but AC power measurement is not audio.

John

But A/D conversion is the same whatever data you are converting. Different applications can have different sensitivities to different kinds of error, but reducing error is always a good idea.

-- Bill Sloman, Nijmegen

If you can't tell a loudspeaker from an electric meter, I'd clearly be wasting my time.

John

speaking of watt maters, are there any sorts of signals that will confuse the old analog meters? I think mine has a stationary current coil and the moving coil+needle are connected to the voltage coil. The wooden case and plated brass connectors are quite lovely.

I think the dual-coil power meters were true RMS, at least at frequencies where the inductance didn't get in the way. High frequencies or pulses might have relative phase shifts in the voltage and current coils, which would make it read low.

If the voltage coil is across the AC line, that's usually pretty much a sine wave, and, if so, high frequency currents have no real power. It might get ugly with both voltage and current nabbed downstream of, say, an SCR phase control dimmer.

John

the

an a

stuff

it

at

ss

t

Quite. But a loudspeaker doesn't do any kind of A/D conversion, so this isn't exactly a useful expansion, although it does give us some kind of insight into your - limited - capacity for systems analysis.

-- Bill Sloman, Nijmegen

stuff

How many electric power meters have you designed? I'm up around 10 or so.

This is the only one we still sell:

They are used mostly in jet engine test cells. United Airlines uses them for testing APUs, too. They are not cheap.

There is still some market remaining for the C180/K20, the utility survey meter, but I licensed that to Enernet for a modest royalty. They know how to sell to utilities.

John

Back around 1980 I was involved with the design of a Solid State Demand=20 Recorder (SSDR). It used an Intel 8085 processor and we did development = on a=20 Millenium 2000 computer system which used CP/M and 8" floppy disks. It = used=20 blue plastic cartridges (the same form factor as 8 track tapes), and = used=20

2716 and 2732 EPROMs which were then replaced with Hitachi 48016 EEPROMs = and=20 then finally SEEQ 2716 EEPROMs. It was designed to read the pulses from = four=20 external wattmeters and recorded the totals in something like 13 bit = words=20 which were squeezed into the memory storage, requiring some digital=20 acrobatics to resolve the data. Memory was still expensive in those = days! We=20 also added a 300 baud modem which was standard technology of the era.

I know that EIL was geared up to make LOTS of these devices, based on = the=20 promises of a design engineer who had basically stolen the technology = from=20 another company which we eventually acquired. But as I remember, he was=20 difficult to work with, especially when he became my engineering = manager. He=20 tried to micromanage the project, and refused to acknowledge the=20 contributions and suggestions of his underlings, and we had a big angry=20 meeting when things came to a head.

By the time we had it working reasonably well, around 1982, technology = had=20 surged ahead, and the Z80 became my preferred processor, and we began to = do=20 development on the new-fangled IBM PCs which used MSDOS. These devices=20 probably were obsolete by the time they were in production, and we = didn't=20 sell anywhere near as many as projected. When the company disbanded, I=20 bought $50,000 of their excess inventory for about $600. I still have = much=20 of it. I could not find anything on these SSDRs online. The closest is = what=20 you link above, and maybe:

And hey, guys, please stop arguing! Sheesh!

Paul=20

quantities for

and even

SPI=20

Nice, but resolution is not accuracy.

?-)

of the

span a

My stuff

as it

that

class

that

on

nal

So you obviously know exactly how to do it, having taken over the world market ...

-- Bill Sloman, Nijmegen

the

span a

stuff

The world market for VME power metering boards, yes. VME is one of our main customer bases.

You may not be familiar with the concept of "customer bases." That's a group of companies who use a technology (VME, in this case) and prefer to buy from us when they can. Selling to utilities is a completely different thing, something we don't care to attempt any more. The ones we've worked with (Niagra Mohawk, NOPSI, various others) were a PITA. The aerospace guys are almost universally great.

John

on of the

and span a

ng. My stuff

ll as it

uce that

12-class

ut that

as on

ournal

be

It was fashionable back in the late 1980's, when Cambridge Insturments introduced the first fully computerised electron microscope , built around a VME back-plane. It was ridiculously expensive, and the next version dumped the back-plane. I'm surprised that there's still a market for VME boards, but then I'm surprised that there's still a market for 555's - physicists don't know much about electronics, and are presumably prone to keep on using obsolete solutions that "just work".

I seem to understand it rather better than you do.

still

The utility companies move enough product to justify penny-pinching detail design. Your forte is in low volume products for specialised markets, where keeping the design costs down justifies expensive choices like persisting with the rather antiquated VME bus.

The last time I needed a backplane, I designed it around LVDS signals on balanced pairs - being worried about radiated interference. We never go to build it.

-- Bill Sloman, Nijmegen

of the

span a

My stuff

as it

that

C12-class

that

journal

VME is roughly a billion dollar business, and has outlasted many architectures that were supposed to kill it. PCI Express was one, and it's already obsolete.

I'd be happy to have 1% of a billion-dollar obsolete market.

Well, I have hundreds of customers. How many do you have?

We're doing PCI Express too, and controllers with one monster FPGA that does most everything, with a Wishbone or Avalon bus inside. Avalon isn't actually a bus, it just pretends to be one.

Of course not.

John

Oops, meant PCI. Compact PCI was supposed to kill VME.

PCI Express still has a few years left.

John

y"

ution of the

cy and span a

ering. My stuff

well as it

reduce that

I C12-class

about that

o was on

a journal

rly be

me

s.

or

The PCI bus wasn't exactly well-designed, as I pointed out here a number of years ago.

I imagine you would. It's still a legacy market, though it will probably outlast you.

None that I know of. What has that got to do with understanding? It establishes that that you do have some minimal grasp of the concept, but the fact that your sales seems to be confined to a limited market that isn't actively expanding would suggest that you do have something to learn.

But PCI is fading, as you point out, and nobody cares what kind of bus you use inside your controllers.

The intended customer wasn't alowed to take on any more graduate students - a fact which didn't have much to do with the technical choices I was making. The guys who had taken him on a customer might have been more careful about doing their homework, but Dutch university politics is pretty much a closed book to me.

-- Bill Sloman, Nijmegen

distribution of the

and span a

metering. My stuff

well as it

reduce that

C12-class

about that

was on

journal

be

What a useless, pompous fathead you are! You are an expert at stuff you've never done, merely on account of your superior intelligence. What a jerk.

Sales are up, maybe more up than I'd prefer. Well, there's not much snow so far, only worth skiing a couple hours in the afternoon, so I may as well design stuff, I guess.

I did manage to finally silence the microwave oven. It went

BEEP BEEP BEEP BEEP BEEP

at ear-splitting levels whenever your coffee was hot or whatever. The hard part is getting the case open without breaking something. Trashing the piezo is easy after that.

Busses inside computers are almost gone. Everything is integrated into the motherboard. PCI Express isn't actually a bus, it's a serial packet protocol.

VME/Versabus was intended to be a "computer" bus, but now it's an I/O bus for systems with thousands of channels.

John

--
     Wrinkle Cream

I have so much to offer,
though the pack runs fast.
I wonder why they don't call.