optical knob encoder: HP/Agilent 33120A freq generator, tuning-knob failure

[Winfield Hill]

We have about 10 Agilent 33120A frequency-generator / synthesizers
in our lab, and a few other similar synthesizers. They feature a
convenient digital frequency and amplitude programming knob that's
based on optical encoders. A pair of left-right pushbuttons sets
which decade you're adjusting with the knob, and over or underflows
operate on the next higher decade. There's a little indent near
the edge of the knob so you can spin it with your fingertip. You
can also use up-down pushbuttons or directly punch in numbers, but
the spinning knob is very convenient. Typically such encoders have
16 or 32 ticks per revolution. A very nice design feature.

The optical shaft encoders were originally made by HP, but now of
course they're made by Avago. Sometimes they call these a digital
potentiometer or a "Panel Mount Optical Rotary Encoder". For
example, the HRPG-AD16 product line, about $34 each at Mouser.

http://www.mouser.com/Search/Refine.aspx?Keyword=HRPG-AD16

http://www.avagotech.com/pages/en/motion_control_encoder_products/rotary_switches/hrpg-ad1616c/

Since they use an optical 2-bit quadrature grey-code (as opposed
to using a pair of switch contacts), they're supposed to be
extremely reliable.

But recently I've encountered two instruments with identical
tuning-knob failures, one in an elegant 33250A 80MHz model
that's on my bench, and the other in one of our standard
33120A 15MHz instruments.

In both cases the knob seems 'stuck" and not to work as you
turn it, except every now and then you'll get one or two
increments or decrements. Totally useless.

I'm wondering if anyone else has encountered this failure?

 

[N_Cook]

We have about 10 Agilent 33120A frequency-generator / synthesizers
in our lab, and a few other similar synthesizers. They feature a
convenient digital frequency and amplitude programming knob that's
based on optical encoders. A pair of left-right pushbuttons sets
which decade you're adjusting with the knob, and over or underflows
operate on the next higher decade. There's a little indent near
the edge of the knob so you can spin it with your fingertip. You
can also use up-down pushbuttons or directly punch in numbers, but
the spinning knob is very convenient. Typically such encoders have
16 or 32 ticks per revolution. A very nice design feature.

The optical shaft encoders were originally made by HP, but now of
course they're made by Avago. Sometimes they call these a digital
potentiometer or a "Panel Mount Optical Rotary Encoder". For
example, the HRPG-AD16 product line, about $34 each at Mouser.

http://www.mouser.com/Search/Refine.aspx?Keyword=HRPG-AD16

http://www.avagotech.com/pages/en/motion_control_encoder_products/rotary_swi
tches/hrpg-ad1616c/

Since they use an optical 2-bit quadrature grey-code (as opposed
to using a pair of switch contacts), they're supposed to be
extremely reliable.

But recently I've encountered two instruments with identical
tuning-knob failures, one in an elegant 33250A 80MHz model
that's on my bench, and the other in one of our standard
33120A 15MHz instruments.

In both cases the knob seems 'stuck" and not to work as you
turn it, except every now and then you'll get one or two
increments or decrements. Totally useless.

I'm wondering if anyone else has encountered this failure?

Its probably patentable so is there a patent outline for them somewhere, to
show what the innards are?

 

[JW]

We have about 10 Agilent 33120A frequency-generator / synthesizers
in our lab, and a few other similar synthesizers. They feature a
convenient digital frequency and amplitude programming knob that's
based on optical encoders. A pair of left-right pushbuttons sets
which decade you're adjusting with the knob, and over or underflows
operate on the next higher decade. There's a little indent near
the edge of the knob so you can spin it with your fingertip. You
can also use up-down pushbuttons or directly punch in numbers, but
the spinning knob is very convenient. Typically such encoders have
16 or 32 ticks per revolution. A very nice design feature.

The optical shaft encoders were originally made by HP, but now of
course they're made by Avago. Sometimes they call these a digital
potentiometer or a "Panel Mount Optical Rotary Encoder". For
example, the HRPG-AD16 product line, about $34 each at Mouser.

http://www.mouser.com/Search/Refine.aspx?Keyword=HRPG-AD16

http://www.avagotech.com/pages/en/motion_control_encoder_products/rotary_switches/hrpg-ad1616c/

Since they use an optical 2-bit quadrature grey-code (as opposed
to using a pair of switch contacts), they're supposed to be
extremely reliable.

But recently I've encountered two instruments with idenical
tuning-knob failures, one in an elegant 33250A 80MHz model
that's on my bench, and the other in one of our standard
33120A 15MHz instruments.

In both cases the knob seems 'stuck" and not to work as you
turn it, except every now and then you'll get one or two
increments or decrements. Totally useless.

I'm wondering if anyone else has encountered this failure?

I have not seen a failure on the 33XXX series of generators, but the part
# is 0960-0892. Looking at the picture on the find-a-part website shows it
as a mechanical encoder.

http://www.home.agilent.com/myagile...&_afrWindowMode=0&_adf.ctrl-state=u4p3ex5iq_4

Which is obsolete and replaced by 0960-2545, also a mechanical one. It
looks like its the same as used on the 65XX and 66XX power supplies. I do
know that the older 65XX and 66XX do use the (obsolete) optical ones. It
is possible to use the mechanical ones to fix the obsolete optical ones
with a bit of hacking, but the mechanical ones need more turns to change
the output of the supplies for a given amount.

Are you sure it's an optical encoder you're looking for?

 

[Winfield Hill]

JW wrote...

Winfield Hill wrote,


I have not seen a failure on the 33XXX series of generators, but
the part # is 0960-0892. Looking at the picture on the find-a-part
website shows it as a mechanical encoder.

http://www.home.agilent.com/myagile...&_afrWindowMode=0&_adf.ctrl-state=u4p3ex5iq_4
Indeed.

Which is obsolete and replaced by 0960-2545, also a mechanical one.
It looks like its the same as used on the 65XX and 66XX power supplies.
I do know that the older 65XX and 66XX do use the (obsolete) optical
ones. It is possible to use the mechanical ones to fix the obsolete
optical ones with a bit of hacking, but the mechanical ones need more
turns to change the output of the supplies for a given amount.

Thanks, great info!

Are you sure it's an optical encoder you're looking for?

No, and as you say, the service manual identifies the mechanical
one as the part, now that I bothered to look. Given the failures,
perhaps being mechanical makes sense. I had just assumed that
because HP long-ago developed and manufactured a fine optical part,
they'd use it in their expensive premium instruments. Heck, I've
long used their expensive optical encoder in my own designs!

The mechanical part looks much smaller than the optical one, which
might not fit as an upgrade replacement. Hmm, their mechanical part
looks a lot like the ones Spehro was selling as excess inventory.

BTW, as you can see at my Mouser link, the HRPG optical encoder
is not obsolete (yet). Hah, it costs $34 instead of $4.23.
Do we get what we pay for?
http://www.home.agilent.com/myagilent/faces/partDetail.jspx?partNumber=0960-2545

 

[Olaf Kaluza]

I'm wondering if anyone else has encountered this failure?

I bought a power supply E3647A and the encoder feels like very bad
quality. My boss bought the same power supply six month ago and the
encoder feels much better.

So I thought the quality of HP goes south and I bought the freq
generator from Rigol. When HP is selling the quality of
Rigol now, I dont like to pay the price for HP anymore!

Olaf

 

[shrtrnd]

It's the 5V 'rice' bulb mounted in the encoder housing.
The bulb needs to be replaced, but you have to be careful.
There are two bulb lengths easily available, with 3 different current ratings.
You have to center the filament exactly in the encoder, or you get intermittent
reading problems.
(I always make sure I've got the same bulb length, and the bulb is pushed down into
it's holder at exactly the same depth as the original).
Good luck

 

[JW]

JW wrote...

Thanks, great info!

You're welcome.

No, and as you say, the service manual identifies the mechanical
one as the part, now that I bothered to look. Given the failures,
perhaps being mechanical makes sense.

Could be, but I've yet to see a mechanical one go south unless there was
some abuse. Even then they are sometimes repairable. The bend over tabs
that hold the encoder together come apart a bit when the shaft is hit at a
right angle, which seems to happen all the time on the power supplies I
mentioned. I've lost count on the number of times I've repaired them.
After repair I solder two 24ga pieces of bus wire across the top from tab
to tab which helps sturdy them up a bit.

I had just assumed that
because HP long-ago developed and manufactured a fine optical part,
they'd use it in their expensive premium instruments. Heck, I've
long used their expensive optical encoder in my own designs!

The mechanical part looks much smaller than the optical one, which
might not fit as an upgrade replacement.

If you *do* have an optical one, it may depend on whether it solders
directly to a PCB, or is panel mounted. If the latter, it probably has a
cable which attaches to a PCB, and you should be able to adapt the
mechanical one. If you go that route, I've found that it is important to
ground the case of the encoder. You'll probably have to fiddle around with
the wires to get the order right. Trial and error...

 

[JW]

The only failed optical encoders I've seen either had a failed
incandescent lamp (Very early type of encoder).

And what a pain in the butt to get to on a 6034A! I wish I could find a
"side firing" LED that would not only fit in that little slot, but also be
the right wavelength for it to work correctly. I've experimented a bit
with this, but never been successful.

 

[Mark Zacharias]

You're welcome.


Could be, but I've yet to see a mechanical one go south unless there was
some abuse. Even then they are sometimes repairable. The bend over tabs
that hold the encoder together come apart a bit when the shaft is hit at a
right angle, which seems to happen all the time on the power supplies I
mentioned. I've lost count on the number of times I've repaired them.
After repair I solder two 24ga pieces of bus wire across the top from tab
to tab which helps sturdy them up a bit.


If you *do* have an optical one, it may depend on whether it solders
directly to a PCB, or is panel mounted. If the latter, it probably has a
cable which attaches to a PCB, and you should be able to adapt the
mechanical one. If you go that route, I've found that it is important to
ground the case of the encoder. You'll probably have to fiddle around with
the wires to get the order right. Trial and error...

I see mechanical ones go bad ALL THE TIME. I even keep a few of the more
common ones in stock. I was going to mention this sooner, but it was said
that there was a pushbutton function on the OP's unit which the encoders I
see don't have, but one of the pics previously posted look like some I
stock (and pay about 1.00 each for). They go bad from tarnished contacts, in
my opinion, but my friend Arfa opines that it is the migration of silicon
lubricant that is the root of the problem.

Mark Z.

 

[N_Cook]

I designed one into a marine radio and regretted it. If it moves, it
breaks. After about 10 years, all the encoders have probably been
replaced at least once. I also have a few ham radios (Icom IC-735)
with genuinely horrible mechanical shaft encoders, but that respond
well to cleaning.


The CUI data sheet shows versions with a push button option.


In my never humble opinion, y'er both right. It's both tarnishing and
the silicon grease that causes problems.
<http://en.wikipedia.org/wiki/Polydimethylsiloxane>
It tends to absorb aromatic hydrocarbon solvents. It also tends to
have residual acetic acid used in the manufacture. Both are very mild
oxidizers that will eventually rot the surface of any metal contacts.
Once pitted, the contact area is reduced. Silicon grease enters the
voids, and since it's an insulator and incompressible, eventually
ruins the electrical connection. Using soft metals, such as PCB
copper, for contacts helps because the wiper scrapes away any surface
pitting. That only works until the copper is worn away. I've seen
encoders with grooves worn into the PCB traces.

In desperation, I found that cleaning off the grease from the copper
with alcohol, and tinning the copper PCB traces, actually works fairly
well. The problem was that when I lifted the soldering iron, there
would always be a small solder points left behind, resulting in
multiple bumps as the encoder rotated. I managed to file or press
down the points, but the bumps are still obvious. One of my radios
was repaired in this manner several years ago and is still functional
(although the bumps are irritating).





--
Jeff Liebermann [email protected]
150 Felker St #D http://www.LearnByDestroying.com
Santa Cruz CA 95060 http://802.11junk.com
Skype: JeffLiebermann AE6KS 831-336-2558

I don't know if its relevant with these encoders but a few times I've found
with preset size pots that appear in audio mixers. The grease hardens and
then lifts the very feeble wipers of the pots. No track wear at all as the
wiper action is so slight , so why the grease ?

 

[Mark Zacharias]

I don't know if its relevant with these encoders but a few times I've
found
with preset size pots that appear in audio mixers. The grease hardens and
then lifts the very feeble wipers of the pots. No track wear at all as the
wiper action is so slight , so why the grease ?

I think the silicon grease is there to provide the "feel" when you turn the
controls.

Mark Z.

 

[N_Cook]

I think the silicon grease is there to provide the "feel" when you turn the
controls.

Mark Z.

I'd not thought of that as a reason but it makes some sort of sense. I
suppose the first pot one ,I cleared the grease out with meths and
reassembled dry, refitted, and returned to owner must be about 5 years ago.
That one and subsequent ones, have not bounced back with worn tracks , so
superfluous as far as electrical function is concerned , IWS. I suppose
someone will now tell me I should have replaced like for like, not going
against the "designer's" wishes.

 

[N_Cook]

Good question. You might need a microscope to see the groves.

My guess(tm) would be to prevent oxidation and prevent the small
amounts of the resistance material from landing across contacts. I
also recall reading that it reduces pot noise improves the "feel" by
adding some drag, but that's mostly on the shaft, not the wiper.

This explains much:
<http://www.nyelubricants.com/lubenotes/Lubenote_Potentiometers.pdf>

Looks like they have different greases for different problems:
<http://www.nyelubricants.com/applications/sensor_potentiometer.shtml>
--
Jeff Liebermann [email protected]
150 Felker St #D http://www.LearnByDestroying.com
Santa Cruz CA 95060 http://802.11junk.com
Skype: JeffLiebermann AE6KS 831-336-2558

Of course being a seller of lubricants there is no tabling of the long-term
performance as far as immunity to hardening. That "velvet feel" becoming
gumming up the works . And pot track lubrication being insufficient to cause
"aquaplaning" and then with aging/ chemical reaction, hardening and forming
a wedging under the wiper and so loosing contact for all or most of the
track run.

A job for tomorrow is to take some measurements of one of those tiny preset
size pots.
The wiper arcs are something like 3 off 10mm long , and section .2x.4mm IIRC
but I will try and measure the contact force , I'm guessing at this stage of
order 10 gm. Once one wiper lifts I imagine its not long before all 3 lift
at some part of their travel.

 

[N_Cook]

A job for tomorrow is to take some measurements of one of those tiny preset
size pots.
The wiper arcs are something like 3 off 10mm long , and section .2x.4mm IIRC
but I will try and measure the contact force , I'm guessing at this stage of
order 10 gm. Once one wiper lifts I imagine its not long before all 3 lift
at some part of their travel.

Alpha type 4D3-11 pot , footprint 9.6x 11.2mm. 2 wiper arcs about 7 and 9 mm
long. Wiper metal 0.1mm thick x 0.3mm.
Could not measure contact force in-situ. With wiper removed from recess, 10
gm would visibly move one wiper relative to the other. No visible track
scoring under a x30 magnification.
Of course , those old enough, would have come across the effect of hardening
grease in cassette tape mechanisms then VCR. Where an arm pivots on an axle
and return action is just a light torsion spring.

 

[N_Cook]

You won't appreciate proper pot lube until after you try to operate a
mobile radio or HT when it's really cold, and find all the controls
stuck in place due to thickening of the lubricant.

Easily dissimilar metals , Al and brass and sliding fit , then the naval
saying.
Cold enough to freeze the (iron canon) balls off a brass monkey (trivet)

 

[Spehro Pefhany]

We have about 10 Agilent 33120A frequency-generator / synthesizers
in our lab, and a few other similar synthesizers.

Hey, Win:-

We had a mechanical encoder failure on a 33220A arb function
generator. It is a known issue with these encoders and Agilent sent a
repair kit gratis (or we could have sent it in). It's an easy fix, so
I elected to do it myself.


Best regards,
Spehro Pefhany

 

[JW]

Hey, Win:-

We had a mechanical encoder failure on a 33220A arb function
generator. It is a known issue with these encoders and Agilent sent a
repair kit gratis (or we could have sent it in). It's an easy fix, so
I elected to do it myself.

Here's the service note:
http://cp.literature.agilent.com/litweb/pdf/33220A-02B.pdf
The expiration for free service or parts is February 1, 2012.