Prev: cmos latchup in mcu
Next: Built-in audio playback module
From: JosephKK on 24 Apr 2010 06:50
On Tue, 20 Apr 2010 04:11:50 -0700, John Larkin
<jjlarkin(a)highNOTlandTHIStechnologyPART.com> wrote:

>On Mon, 19 Apr 2010 20:50:35 -0700, Tim Wescott <tim(a)seemywebsite.now>
>wrote:
>
>>John Larkin wrote:
>>> Can anybody suggest a good one?
>>
>>Ha ha ha ha ha he he he he he ho ho ho ho ho aaaaaaaaaaaaaaaaagh!
>>
>>(I'm not making fun of you here, that's hysterical laughter).
>
>There seems to be some stuff from, say, 1955 or so. I don't suppose
>much has changed.
>
>>
>>Call Renco Encoders and see if you can sweet-talk them out of a copy of
>>"Feedback Devices in Motion Control Systems", by Robert M. Setbacken.
>>
>>Then call Moog Components -- wade through the Moog website and find the
>>guys that sell the precision industrial and aerospace encoders. Sweet
>>talk them, too.
>>
>>Be aware that all the electronics whizzes who worked on encoders have
>>died of old age -- it's all mechanical engineers; they know bearings,
>>they know winding machines, and they know how to test things, but when
>>you start asking questions about impedance vs. frequency and other
>>seemingly obvious things the best you'll get is a friendly shrug.
>
>Seems that way. A lot of the books and lit seem cartoonish.
>
>>
>>They're variable transformers. They're really inefficient. Their
>>impedance is pretty close to the wiring resistance plus the inductance
>>times radian frequency (no surprise there). They work over an
>>astonishingly large frequency range, although they are traditionally
>>only specified at the frequency that the first customer wanted to use.
>>The drive amplitude is specified in voltage, although if you read
>>between the lines they're limiting the I^2R losses in the primary. Try
>>to ask the guys who design them these days and at best you'll get a
>>friendly shrug...
>>
>>You can order them specified for other frequencies, or sizes, for "some"
>>NRE. "Some" is relative, and Moog is an aerospace company...
>>
>>How tightly are you gonna simulate the resolvers? Just behavioral?
>>1st-order with inductances and winding resistance? Parallel
>>capacitance? Electrical nonlinear effects? Spatial nonlinear effects?
>> Are you gonna simulate multi-speed resolvers?
>
>My gadget is going to be transformers, adcs, dacs, and an FPGA.
>Whatever I can do with that. I'll be generating low-power signals into
>an electronics box that thinks it's connected to an LVDT or a synchro.
>It's representing mechanical positions of things like flaps or bits of
>engines, so it doesn't need arc-second accuracy. Stuff like
>multi-speed would be easy.
>
Multispeed easy?? Not going to happen. Do you know how to
elecromechanically zero a multispeed synchro?
>>
>>There's probably a market for a general-purpose resolver simulator, and
>>possibly even more so for a general-purpose resolver _meter_ (I recall
>>looking and not finding one, at a time that one was desperately needed).
>
>The meters are around.
>
>http://www.naii.com/Angle-Position-Indicators/PSC8
>
>http://www.controlsciences.com/
>
>http://peakelectronics.com/index.php?option=com_content&task=view&id=36&Itemid=62
>
I can see where you think you can break into a lucrative, but niche,
market. I worked on synchro/ digital and digital/synchro converters in
the 1970s (tech days, but i did learn them right).
>
>I'll have to get one if I go ahead with this project. I already have a
>couple of those old 6-dial variable-ratio transformer boxes, the
>ppm-accurate things that last forever... left over from an LVDT
>project.
>
>>
>>But there's lots of metrology issues to overcome if you want the meter
>>or simulator to exceed the accuracy that you can get from a resolver --
>>the really good ones work down to arc seconds, which is "don't breath"
>>territory if you want to check out the physical system.
>
>16 bit electronics should work. There are lots of affordable sub-MHz
>16-bit ADCs and DACs around nowadays.
>
>John
From: John Larkin on 24 Apr 2010 10:50
On Sat, 24 Apr 2010 03:32:44 -0700,
"JosephKK"<quiettechblue(a)yahoo.com> wrote:

>On Mon, 19 Apr 2010 17:31:31 -0700, John Larkin
><jjlarkin(a)highNOTlandTHIStechnologyPART.com> wrote:
>
>>On Mon, 19 Apr 2010 17:24:14 -0700, VWWall <vwall(a)large.invalid>
>>wrote:
>>
>>>John Larkin wrote:
>>>> Can anybody suggest a good one?
>>>>
>>>For their use in old time systems, Rad Lab Series Vol.21.
>>
>>I have that. And most of the other books don't seem to be much newer!
>>
>>>
>>>If you need to connect between a synchro and a resolver, use a Scott-T
>>>transformer. :-)
>>
>>An expensive transformer seems to be a silly way to do a little trig.
>>Adding another multiply to an ARM program costs ... calculates
>>furiously ... $0.00.
>>
>>I'm contemplating designing an LVDT/synchro/resolver simulator. The
>>sensible way seems to be to digitize the excitation voltage and make
>>the various winding voltages with DACs, and do all the rest in a uP or
>>FPGA.
>>
>>John
>>
>
>I am presuming that you have the wit to digitize ratiometrically between
>the excitation and the output(s). Then the fancy transformers are
>unnecessary.

I have not the wit to understand what you're talking about. The real
point being, do you?

John

From: John Larkin on 24 Apr 2010 10:56
On Sat, 24 Apr 2010 03:50:50 -0700,
"JosephKK"<quiettechblue(a)yahoo.com> wrote:

>On Tue, 20 Apr 2010 04:11:50 -0700, John Larkin
><jjlarkin(a)highNOTlandTHIStechnologyPART.com> wrote:
>
>>On Mon, 19 Apr 2010 20:50:35 -0700, Tim Wescott <tim(a)seemywebsite.now>
>>wrote:
>>
>>>John Larkin wrote:
>>>> Can anybody suggest a good one?
>>>
>>>Ha ha ha ha ha he he he he he ho ho ho ho ho aaaaaaaaaaaaaaaaagh!
>>>
>>>(I'm not making fun of you here, that's hysterical laughter).
>>
>>There seems to be some stuff from, say, 1955 or so. I don't suppose
>>much has changed.
>>
>>>
>>>Call Renco Encoders and see if you can sweet-talk them out of a copy of
>>>"Feedback Devices in Motion Control Systems", by Robert M. Setbacken.
>>>
>>>Then call Moog Components -- wade through the Moog website and find the
>>>guys that sell the precision industrial and aerospace encoders. Sweet
>>>talk them, too.
>>>
>>>Be aware that all the electronics whizzes who worked on encoders have
>>>died of old age -- it's all mechanical engineers; they know bearings,
>>>they know winding machines, and they know how to test things, but when
>>>you start asking questions about impedance vs. frequency and other
>>>seemingly obvious things the best you'll get is a friendly shrug.
>>
>>Seems that way. A lot of the books and lit seem cartoonish.
>>
>>>
>>>They're variable transformers. They're really inefficient. Their
>>>impedance is pretty close to the wiring resistance plus the inductance
>>>times radian frequency (no surprise there). They work over an
>>>astonishingly large frequency range, although they are traditionally
>>>only specified at the frequency that the first customer wanted to use.
>>>The drive amplitude is specified in voltage, although if you read
>>>between the lines they're limiting the I^2R losses in the primary. Try
>>>to ask the guys who design them these days and at best you'll get a
>>>friendly shrug...
>>>
>>>You can order them specified for other frequencies, or sizes, for "some"
>>>NRE. "Some" is relative, and Moog is an aerospace company...
>>>
>>>How tightly are you gonna simulate the resolvers? Just behavioral?
>>>1st-order with inductances and winding resistance? Parallel
>>>capacitance? Electrical nonlinear effects? Spatial nonlinear effects?
>>> Are you gonna simulate multi-speed resolvers?
>>
>>My gadget is going to be transformers, adcs, dacs, and an FPGA.
>>Whatever I can do with that. I'll be generating low-power signals into
>>an electronics box that thinks it's connected to an LVDT or a synchro.
>>It's representing mechanical positions of things like flaps or bits of
>>engines, so it doesn't need arc-second accuracy. Stuff like
>>multi-speed would be easy.
>>
>Multispeed easy?? Not going to happen. Do you know how to
>elecromechanically zero a multispeed synchro?
>>>
>>>There's probably a market for a general-purpose resolver simulator, and
>>>possibly even more so for a general-purpose resolver _meter_ (I recall
>>>looking and not finding one, at a time that one was desperately needed).
>>
>>The meters are around.
>>
>>http://www.naii.com/Angle-Position-Indicators/PSC8
>>
>>http://www.controlsciences.com/
>>
>>http://peakelectronics.com/index.php?option=com_content&task=view&id=36&Itemid=62
>>
>I can see where you think you can break into a lucrative, but niche,
>market.

I don't have to break into a market. A pretty big aerospace company is
asking me to do this to replace a board that they designed themselves
and don't want to/can't make any more.

A lot of big aerospace companies and national labs used to design
their own electronics. Sometimes it was good, but often not. The trend
is now towards outsourcing, for several reasons.

John

From: VWWall on 24 Apr 2010 19:37
John Larkin wrote:
> On Sat, 24 Apr 2010 03:32:44 -0700,
> "JosephKK"<quiettechblue(a)yahoo.com> wrote:
>
>> On Mon, 19 Apr 2010 17:31:31 -0700, John Larkin
>> <jjlarkin(a)highNOTlandTHIStechnologyPART.com> wrote:
>>
>>> On Mon, 19 Apr 2010 17:24:14 -0700, VWWall <vwall(a)large.invalid>
>>> wrote:
>>>
>>>> John Larkin wrote:
>>>>> Can anybody suggest a good one?
>>>>>
>>>> For their use in old time systems, Rad Lab Series Vol.21.
>>> I have that. And most of the other books don't seem to be much newer!
>>>
>>>> If you need to connect between a synchro and a resolver, use a Scott-T
>>>> transformer. :-)

I meant this in jest, referring to methods as old as the reference.

>>> An expensive transformer seems to be a silly way to do a little trig.
>>> Adding another multiply to an ARM program costs ... calculates
>>> furiously ... $0.00.
>>>
The Scott-Tee was used to convert from three synchro windings to two
resolver windings. No trig was involved, just a two phase/three phase
conversion, which the power types used when two phase power was around.

For use in a synchro/resolver system/simulation the "expensive
transformer" can be replaced by two op-amps and a hand full of
resistors. The resistors replace the ~86% tap on one transformer.

>>> I'm contemplating designing an LVDT/synchro/resolver simulator. The
>>> sensible way seems to be to digitize the excitation voltage and make
>>> the various winding voltages with DACs, and do all the rest in a uP or
>>> FPGA.
>>>
>>> John
>>>
>> I am presuming that you have the wit to digitize ratiometrically between
>> the excitation and the output(s). Then the fancy transformers are
>> unnecessary.
>
They never were.

> I have not the wit to understand what you're talking about. The real
> point being, do you?
>
> John
>
Virg Wall, P.E.
From: Jim Thompson on 24 Apr 2010 20:32
On Sat, 24 Apr 2010 16:37:55 -0700, VWWall <vwall(a)large.invalid>
wrote:

>John Larkin wrote:
>> On Sat, 24 Apr 2010 03:32:44 -0700,
>> "JosephKK"<quiettechblue(a)yahoo.com> wrote:
>>
>>> On Mon, 19 Apr 2010 17:31:31 -0700, John Larkin
>>> <jjlarkin(a)highNOTlandTHIStechnologyPART.com> wrote:
>>>
>>>> On Mon, 19 Apr 2010 17:24:14 -0700, VWWall <vwall(a)large.invalid>
>>>> wrote:
>>>>
>>>>> John Larkin wrote:
>>>>>> Can anybody suggest a good one?
>>>>>>
>>>>> For their use in old time systems, Rad Lab Series Vol.21.
>>>> I have that. And most of the other books don't seem to be much newer!
>>>>
>>>>> If you need to connect between a synchro and a resolver, use a Scott-T
>>>>> transformer. :-)
>
>I meant this in jest, referring to methods as old as the reference.
>
>>>> An expensive transformer seems to be a silly way to do a little trig.
>>>> Adding another multiply to an ARM program costs ... calculates
>>>> furiously ... $0.00.
>>>>
>The Scott-Tee was used to convert from three synchro windings to two
>resolver windings. No trig was involved, just a two phase/three phase
>conversion, which the power types used when two phase power was around.
>
>For use in a synchro/resolver system/simulation the "expensive
>transformer" can be replaced by two op-amps and a hand full of
>resistors. The resistors replace the ~86% tap on one transformer.
>
>>>> I'm contemplating designing an LVDT/synchro/resolver simulator. The
>>>> sensible way seems to be to digitize the excitation voltage and make
>>>> the various winding voltages with DACs, and do all the rest in a uP or
>>>> FPGA.
>>>>
>>>> John
>>>>
>>> I am presuming that you have the wit to digitize ratiometrically between
>>> the excitation and the output(s). Then the fancy transformers are
>>> unnecessary.
>>
>They never were.
>
>> I have not the wit to understand what you're talking about. The real
>> point being, do you?
>>
>> John
>>
>Virg Wall, P.E.

My apologies, Virg, Larkin is a bit like Al Gore :-)

...Jim Thompson
--
| James E.Thompson, CTO | mens |
| Analog Innovations, Inc. | et |
| Analog/Mixed-Signal ASIC's and Discrete Systems | manus |
| Phoenix, Arizona 85048 Skype: Contacts Only | |
| Voice:(480)460-2350 Fax: Available upon request | Brass Rat |
| E-mail Icon at http://www.analog-innovations.com | 1962 |

The only thing bipartisan in this country is hypocrisy
First  |  Prev  |  Next  |  Last
Pages: 1 2 3 4 5
Prev: cmos latchup in mcu
Next: Built-in audio playback module