The biggest problem with OCR especially with manuals is that The OCR engine unlike we humans is not context sensitive,? for example often"I" ,"l" and "1"? (pronounced , eye , el and one) are confused by the OCR engine ("B" and "8: are the next candidates). This is particularly problematic on Schematics and especially HP parts lists which were often printed with a tiny 6pt font. OCR engines are designed with text dialogue and news paper column in mind. Schematics are also a big problem as a result because the OCR engine doesn't always know what to do with R101 ( was that an "O" or a "0" ) hanging out in space and not part of sentence. You and I reading it know that is R101 and not RLOL because we read it in context. Hp at one point made it even worse as the font they used was identical for lower case "L" and the number "1"

Dave
ArtekManuals

--
Dave
Manuals@...
www.ArtekManuals.com

开云体育

Hi David,

Absolutely correct and I'd certainly persist with OCR if I was getting good results. OCR with newer manuals may be less of a problem, but the only scanner I own, a HP OfficeJet Pro 8710, really struggles to recognise text from older, yellowed, faded, and well thumbed pages.?The OCR on that device always needs plenty of manual correction, especially with data presented in tables, captions and footers, and it's not always happy working with text layed out in two columns.

Maybe I'm not using the OCR correctly, but the time taken to scan and then reassemble the jigsaw of OCR always seems more than the time taken just typing at 80 words per minute.

To all of you getting good results with OCR: I am insanely jealous!

Kindest and warm regards,
Terry

开云体育

Hi Dave,

Sorry for the late reply.

"Any idea if that manual has been scanned? A quick Google doesn't turn it up anywhere."

I've found no scanned copies of the 2410B manual looking around the usual places on the net, which leaves few options.

1) There are some service manual reproduction companies with proper copyright permission that are open to requests of what other manuals they should offer. I used one such company today, ArtekManuals.com. For $12.95 I got a scanned copy of a rare HP 5376A Programmer manual for the HP 5360A computing counter. In this case, I was wondering what it would take to substitute a small single board computer, such as a Raspberry PI 4, in place of the ultra rare HP 5376A programer. As expected, it's a fascinating document, but I'll have to throw many many hours at it just to plan a result. I'll ask ArtekManuals if they will scan a?2410B manual and get back to you with a PM.

2) If you haven't tried it, there is a group called?[manualexchange] where you could ask.

3) If your prepared to wait till about late Jan or early Feb, I'll e-mail you a copy of my yet-to-be scanned manual. When scanning a manual, I usually re-type the text, instead of just imaging the whole page, as this allows the reader to do a word search. Images of circuit diagrams etc are still scanned from the original manual. Images of circuit boards don't always scan well, so I substitute high resolution colour photos if I have the actual boards. But the goal is always to have a verbatim copy that's indistinguishable from the original in fonts, layouts and page count.?

Kind regards,
Terry.

Hi all,

E18 has gone once I changed the DAC converter, the one for fine-tuning the first LO. I realigned the unit and everything looks fine - I just cannot use it since I do not have a noise source... even a cheap one.

I only managed to get an SMD-Version of the DAC, so I had to do some artistical wiring work to put it in, see the picture attached. At the end, to get the board back in place, I had to take off one of the rubber bumpers on the Isolator - it was just underneath the DAC and probably touching the original one.

cheers, merry Christmas, happy new Year!
Martin

Hello,

you also need to keep one more thing in mind which is bad for these seeking a cheap 3458A.

If you take a completely beat up one, and a decent lawyer, Keysight will - AFAIK - repair it for a reasonable set fee. This essentialy gives a new multimeter and puts a very high minimal bound on sales prices. If I would find a 3458A in the bin tomorrow and it was shot up with a Gepard rifle and was urinated on by both fox and female, as long as I can find all the parts, I would try to get Yaniv to force KS to honor its repair policy and flog the result on ebay.


Something I have had a lot of fun with is SolarTron. But I am not a big voltnut, my 7061 recently failed and I dont miss it so my wife balks heavily at the cal and repair cost.


Tam



With best regards
Tam HANNA

Enjoy electronics? Join 15k7 other followers by visiting the Crazy Electronics Lab at

My reaction when I read the subject title was, "Yeah, you and a few hundred others."

The recent introduction of the RoHS "black" edition indicates that Keysight isn't planning to replace the design. Old designs which are still in production hold their value well. The 34401A continued in production for an unusually long time because it was stipulated in so many test equipment lists in service manuals.

I picked up a couple of 34401As via ebay for ~$250 each. One was wonky. The cause was solder flux from an old repair. I cleaned it with isopropyl and it is the best of the pair. That's about 25% of the cost of the new equivalent. There are several on ebay now which will sell in the $200-300 price range. The same valuation puts an easily repairable 3458A at $2350. And an incomplete unit whether missing or unrepairable board at ~$1000-1500 as a parts mule. Provided it's not a set of known bad boards.

A working 3458A is worth $3000-5000. With RoHS compliant models available, there should be a drop in price on older units. But not likely to be huge.

Volt nutting has become rather fashionable.

Have Fun!
Reg

Thanks Dan. Sounds like a good place to begin. I have an Anritsu VNA ... just need to figure out how to use it for this.

I ran the unit over night and had my SA (4407B) on MAX HOLD as I was concerned about frequency drift. Set on CW, 4.5 GHz, 0 dBm:


If I am reading the specs correctly (stability) 250 kHz seems high. I can see the drift in real time but it's quite slow.

Mark

On Mon, 16 Dec 2019, victor.silva via Groups.Io wrote:

Tried to send direct without looking at the header so it bounced... Sorry
for posting to the list but it won't go otherwise. Here is the original
message:

=== Cut ===
It looks like Paul Bicknell asked for that one first but if he didn't buy it
you can get it. I took some pictures minutes ago, they are available at



It is a 44+ MBytes archive with several photos.
=== Cut ===

Sergey,

I would be interested in the 83570A.

I agree with your recommendation for removing the battery.? I did that to
mine 8 years ago and have not regretted it.

--Victor

---
*
* KSI@home KOI8 Net < > The impossible we do immediately. *
* Las Vegas NV, USA < > Miracles require 24-hour notice. *
*

Hi. GOT IT!
There was a ground loop gathering noise and hum.?
I moved the ground connection of the third gang of the variabile Cap in the same point where Wien bridge resistive leg is grounded.
Now I got -50,1 dB THD @ 1 KHz, -10 dB under spec.
I'm really happy.
Thank you!

Hello,

Yes, this is the input receiver section where RF is down converted to somewhere around 25 - 270 MHz to fit the second mixer, which brings it to exactly 2.5 MHz.
I have tried to feed the output of the PA through an attenuator, but it does appear as if the mixer needs full power. I have attached a picture of the board and a measurement of saturated output (input power of 5 dBm, seems like the level from the VNA also). The instrument seems to lock for any LO setting below approximately - say - 520 MHz so the following mixer appears to be working. The bad thing is that during a sweep, the higher LO frequencies appear and then the VNA gets stuck. Needless to say, the higher RF makes more use of higher LO frequencies...

Anyone knows of a good amp that provides 30 dBm _flat_ between 360 and 540 MHz??

Or, any suggestions on how to trim the current amp to give those final dBm at the higher range?

Perhaps the step recovery diode isn't properly matched right now - there is a PI net of caps and a trim pot at the output. They are set as they were from the beginning. The problem is however that it is almost impossible to trim the unit when it is installed (the lid is holding the amp). My guess is that a generic tuning is sufficient and it seems like the output power curve well matches how well the mixer works.

It's great to see that it now locks also at high RF, but frustrating that it won't run a full sweep. Should perhaps rewrite the code to only use frequencies below 500 MHz?... Not likely...

Sad thing with Wiltron is that any technical documentation at the level where HP provided is nearly nonexistent. PLEASE anyone, PROVE ME WRONG!

Regards,
? Staffan

Sergey,

I would be interested in the 83570A.

I agree with your recommendation for removing the battery.? I did that to mine 8 years ago and have not regretted it.

--Victor

I agree with that. A few years ago 3458As with issue would go for $1500, I even saw some under a thousand.
Lately they are all over $2000, for example the unit mentioned on ebay in this thread, went for?$2,222.22.
It was an older unit, but did have option 002 (and 001).

The problem with trying to get a dead but complete unit is that a dead unit draws a lot of attention.
P/S issues are usually the easiest to diagnose and fix (IMHO), so I think those units draw attention from people hoping for an easy score.

--Victor

Ok, I think I can answer my first question. The CRC with additional Rs to ground is DC blocking capacitors either side of a pi-pad attenuator. Given the 300 and 18 ohm resistance, they're just about 3dB with 50Ohm impedance. This means that the input LO has 2x20dB of gain from the SBB5089Z as well as 3x-3dB of attenuation from the pi-pad attenuators, so overall 31dB of gain, lifting my input LO from -12 to +19dB into the mixer (ignoring other insertion losses). The absolute maximum rating for the mixer is 13dBm for the RF in, so this seems plenty of power to work with the mixer. I will check what level the voltage controlled attenuator is being driven to.

Happy to hear suggestions on the RF side, my RF knowledge is still quite weak.

On 12/16/19 3:40 AM, Terry Gains wrote:

I have a old, much-used 2410B manual to work from, but will get a new
one to make the job easier.

Any idea if that manual has been scanned? A quick Google doesn't turn
it up anywhere.

-Dave

--
Dave McGuire, AK4HZ
New Kensington, PA

Out of curiosity quick scan on ebay:

1 pc NOT working HP 3458A Multi-meter "For parts or not working"

Already at $2.132, 12 minutes to close.



Not related in any way to the auction.

As promised, here's an update on adding a LO port and external tracking generator for the 8591A.

The output of the (originally terminated) TG LO port on the first mixer is about -12dB once I routed it out of the back of the unit, but its exactly in the right frequency range.

I bought the BG7TBL 2017-11-26 tracking generator, connected it but its not working. Borrowing a second spectrum analyser, I can see that the TG signal is 400 MHz lower than the monitored frequency of the 8591A, so its working but the frequency of the LO in the TG is 400MHz is too low. No problem, that's what the front "Freq. Adj." is for right? Wrong, it only allows for a change in frequency of about 30kHz, so its for fine tuning only. I check my order, and it turns out I bought the TG for the Anritsu MS610 as it saved me about ?20, and sure enough, the difference between the MS610 and the 8591A first converter IFs is 400MHz. That ?20 is starting to look worthwhile now.

I also note that the max output of the TG is -20dB. Again, there's a "Lev. Adj." dial on the front, but this only changes it from -40dB to -20dB and at max there's a lot of spurs at the max setting. Right now, it looks like the LO from the 8591A needs to be amplified before heading into the TG?

Time to tear the TG down and see how I can adapt it to work for the 8591A by changing the oscillator frequency. Here's a picture of the insides:


There's an ATMega8 which seems to control the PLL hidden under the small RF shield, while the actual mixing etc seems to be under the larger shield. Its well laid out but the front pots/knobs are really cheap. Its a two layer board too, but the underside is nearly an unbroken ground plane. The only exception are some very short jumpers to the other side underneath the RF shields, mainly for the digital signal lines. Soldering quality seems good with some small bodge jobs here and there. The large RF shield only had one soldered tab and the others were just bent so it was easy to remove. The PLL shield was a horror to remove.?
?
The LO input comes through two stages of amplification using two (+20dbm 0-4GHz), before hitting the ?(1.5-4.5GHz RF/LO input, DC-1.5GHz output), so its specified output freq. range is a little below the range of the 8591A. This seems to be plenty of gain for the LO input, so maybe there's some attenuation to provide isolation? The PLL is a (137.5 MHz-4.4GHz), driven by a three wire SPI-like interface from the ATmega8a in the bottom right. This is then attenuated using a voltage-controlled?attenuator before again being buffered by an SBB5089Z before entering the . After the mixer there's two unknown chips (filters?) labelled F1 and F2, before another SBB5089Z, then some kind of detector(?)/power-splitter before entering into the LM358 op amp, which I think must be comparing against the potentiometer as its output goes into the control pin of the AT-108 attenuator, thus I'm guessing its the level control.

The on-board pots are "overridden" by the front panel pots and do nothing while they are attached so I think the board is designed to run stand-alone from the case. Strange that they're populated anyway. There's no serial output from the ATMega, despite the on-board RX/TX connector in the bottom right, but I'm guessing this is how the PLL frequency is varied/programmed between the different units sold.?

My next step is to trace out the ATmega8a connections to see if its just controlling the PLL, then record the exact commands sent to the PLL, then reprogram the ATmega8 to send a command with a frequency 400MHz higher. If this works, I'll look at adding a display interface to the ATmega to allow a controllable frequency offset for offset sweeps and maybe other fun ideas.

Let me know if you want any more information on this tracking generator, I'll update again once my AVR programmer has arrived and I've scoped out everything that its doing.

I have two questions. Between the SBB5089Z there's often a CRC component with R=18Ohms, and two 300Ohm grounding resistors either side of it. I understand the need for the C, as its blocking the DC bias for the SBB5089Z, but not why the resistors are there. Is it for attenuation to provide isolation? Its not a matching circuit as everything is internally matched to 50Ohm. There's even a case where its not used, i.e. from the PLL to the mixer. The second question is any ideas why the two stages of +20dB gain each are not enough for the input LO? Is it because those resistors are indeed there for attenuation? I've bought a cheap RF amplifier module off amazon, I'm hoping just adding this to the LO input will bring up the amplitude of the TG output.
Thanks!?

I guess whichever of us gets the time to do the boards will then have to share with the other.??

You bet. There are many crowds that can make the PCBs with cut outs and edge connectors etc. For example the work could be sent to?although they will probably squawk when I ask for "Gold PCB tracks". :)

I have a old, much-used 2410B manual to work from, but will get a new one to make the job easier.

Kind regards,
Terry