Owners of a Hameg HM103-1 or -3 oscilloscope: can you help me?

[Greybeard]

Hi,
I recently managed to snag this analog gem at a flea market for a good price.



(c) Gertjan 

Unfortunately, the sensitivity of the Y amplifier is much lower than normal.
When I connect the 200 mVpp cal-square-signal  via a 10:1 probe, I only see a straight line on the screen (even with the switch in the 5 mV/div position).
Connected directly via a short piece of wire, I can at least see a peak-to-peak voltage of almost one Y grid unit on the screen, so slightly less than 5 mVpp.
I checked the cal signal with another oscilloscope and also checked the internal supply voltages - everything was fine.

When I switch to the component tester (CT), I see a horizontal line about 8 grid units long when the CT jacks are open.
When the CT jacks are shorted, I see a vertical line about 7 grid units long. So far, everything's fine.
So, the Y-final-stage from the CT switch onward is fine, and the problem must be in the Y-preamp or attenuator.

I've spent several hours trying to systematically measure and track the signal.
I also replaced the input JFET pair U440, the CA3046, and the uA733 with working parts (2x BF245 roughly selected, CA3086, LM733), unfortunately without success.

I then simulated the circuit in LT-Spice (using generic transistors) to determine the signal amplitudes at relevant points, and the simulation worked.
Compared to reality, however, the amplitudes from the last two attenuation stages onward are already significantly lower than simulated.

Therefore, I now need real-world comparison values ​​from a friendly HM103-1 or HM103-3 owner (the HM103-2 has a completely different circuit).
I think the following values ​​would help me:

• First, connect a sine generator with 100 mVpp output amplitude to the Y input (without a 10:1 probe divider).
  
• Then switch to the attenuator position 5 mV/div.

Code: [Select]

               | Voltage p-p | Voltage RMS (DC) |
uA733 Pin 1    |             |                  |
uA733 Pin 14   |             |                  |

Please use my attached schematics, as the original had some incorrect pin numbering.

The red numbers I added to the PCB layout are the connection numbers of the attenuator section between the preamp and the intermediate amp.

EDIT: I have reduced the wanted measuring points somewhat, since I was able to narrow down the error to the area before the 733.

[moffy]

VR103 on the ua733 sets the differential gain of the 733 anywhere up to 400 for a short, check the bias on its input and output that they are as expected.

[Greybeard]

I was able to narrow down the error to the area before the 733.
The two input signals of the 733 are in phase (In2: 70mVpp, In1:30mVpp).
I think In1 should be a DC level without overlayed AC.
I'll post my measurements later...

[moffy]

It looks like it has a couple of unity gain buffers composed each of an NJFET and NPN current source which is then buffered by an emitter follower each that feeds a resistive divider. The top buffer is where the signal comes in and the bottom buffer doesn't appear to have any signal but is there for biasing/symmetry reasons. Some signal might be impressed upon Pin14 of the UA733 depending on the resistive divider position, I don't think so, but I'm not sure how it switches. Pin 1 of the ua733 is where the signal comes in, because of their open loop nature the ua733 has a tendancy to saturate its output if the input bias is not carefully balanced.

[Greybeard]

Update:
I've detected the first DC voltage difference at the source of the JFETs: 3.4 V instead of simulated 1.8 V.
I suspect this is a series variation of the U440 dual JFET (symmetrical values).
Therefore, the DC voltage values ​​in the subsequent signal path are also shifted.
The measured AC voltages are as expected, with the exception of Vin1 (30 mVpp) and Vin2 (70 mVpp).
Both input signals of the differential amplifier are in phase!
I think Vin1 should be 0 mVpp (= DC voltage for temperature compensation) and Vin2 should be approximately 150 mVpp.
The AC output voltages Vout1 and Vout2 are much too low (120 mVpp); I was expecting about 3 Vpp.

An interesting point:
When I remove the 733, Vin1 and Vin2 are as expected (Vin1 = 0 mVpp and Vin2 = 150 mVpp).
I suspect that the AC signal at Vin1 is coming from Vin2 via the 733 (is that possible?)...

I checked it with another 733 from my parts stash, but the behavior is the same.
I suspect the problem is with the attenuator.

I tried measuring the attenuator resistances and therefore removed the CA3046 and the 733.

Measured values:
Pin 2-Pin 4: 274 ohms -> OK
Pin 1-Pin 3: 284 ohms (calculated: 284 ohms) -> OK
Pin 1-Pin 2: 1.36 kOhms (calculated: 1.39 kOhms) -> OK
Pin 2-Pin 3: 1.40 ohms (calculated: 1.43 kOhms) -> OK

What's going on here?

[moffy]

The input voltage range for the ua733 is +/- 1.0V and you are measuring +1.5V probably means the input is saturated.

[Greybeard]

The input voltage range for the ua733 is +/- 1.0V and you are measuring +1.5V probably means the input is saturated.

You're so right! 

I've now replaced the input JFET for the second time with another U440, which I removed from an HM203 oscilloscope.
This time, however, I double-checked that the donor device was working.
I think I replaced it with a defective/wrong JFETs the first time.

This time, the input voltages of the uA733 are around 0.7 V (both).
The oscilloscope is working normally (of course, I'll still have to do some calibration).,

Sometimes you just have to trust your first instinct and double-check everything anyway. 

But I'm very surprised that the defective dual JFET is wonderfully "symmetrically defective" and the input signal still came through (AC-wise).
Isn't that strange?

[moffy]

Always glad to hear when there is a good outcome, congrats.

[Greybeard]

Do you know of a cheap replacement for the discontinued U440?
I don't want to spend more than the entire scope cost me at the flea market.
Perhaps there are suitable paired JFETs available as current SMDs or cheap leaded single JFETs that I can glue together?
Which parameters are important to match in this application?

[moffy]

They are reasonably wideband unity gain buffers, they need to be matched in a single package to maintain bias levels, whatever you choose you are going to have to adjust the current sources to get the bias level you require, even new U440 would need the currents adjusted because of the wide variation in Vgsoff from -1V to -6V.

[Gertjan]

Hi Greybeard,

I was a bit surprised to find one of my HM103 photos in your opening post .

So I presume you saw my HM103 repair report on Circuits Online.
This topic is in unfortunately in Dutch, but with many photos.
Mine had multiple problems, mostly small ones, and mostly due to old age and a lot of usage. Good thing is that Hameg used standard components. So easy to replace.

Do you know of a cheap replacement for the discontinued U440?
I don't want to spend more than the entire scope cost me at the flea market.

I replaced the U440 with a pair of matched BF245B. One half of the U440 was still OK, and I looked for the best matching FET (in my stock) to this half U440.
Most important in matching the 2 replacement FETs is searching for the best match between both FETs (Id/Vds and Id/Vgs curves), for best DC stability. (more info in the Circuits Online topic)


HM103-oude-en-nieuwe-dubbelFET-2000pix.jpg

With this matched BF245B pair I measured a pre-amp bandwidth of 60MHz, and DC stability was adequate. Should be fine for a 10MHz scope .


HM103-Y-versterker-gereviseerd-2000pix.jpg

However, the U440 is obsolete, but recently I found that it is for sale at the moment .
I just bought 2 at Display Elektronica. Mine turned out to be real old stock Siliconics, from 1988 and 1991 .
(Display is a shop that buys the stock of other shops that are closing down)

regards, Gertjan.

[Greybeard]

You caught me, I actually borrowed the photo from your website (please forgive me for that).
The photo was so good that I saved myself the trouble of photographing my device. The devices are otherwise identical.
Of course, I had already read your article "repair from hell" (Google translated, some parts are quite funny).
Your preamplifier also has this ominous 4.75kOhm resistor, which was subsequently mounted "in the air wiring."
At first, I assumed that a hobbyist had added it (my device still has some other tinkering traces from a previous owner), but since I've now seen the resistor in other pictures, it seems that Hameg shipped it that way after all.

Could you please do me a favor and measure the voltages at pins 1 and 14 of the uA733 (Y input = GND)?

I was curious and put my two U440s on the curve tracer, and surprisingly the "defective" one isn't defective at all - the curves are just quite shifted (aging, series variation...?).



(These are two different U440s. The individual pairs match perfectly)

If I compare my measurings to the Vishay datasheet...



... I come to the conclusion that both of my U440s are in spec!

I'm now wondering whether I can adapt the circuit to it.
I probably just need to replace a few resistors (but which ones?). R114/R115?

EDIT: added datasheet

[Gertjan]

Hi Greybeard,

You are forgiven for the photo .

I do not quite understand your U440 measurement. The U440 is a matched pair of JFETs mounted in a single package.
So are the red and green measurements of one FET in two different U440's? Or are red and green the 2 FETs in one U440?

I found that a variation in gain / transconductance is not very critical. If you measured 2 U440's I would expect both of them to work. (presuming a good match between the FETs in the U440)
But the match between the 2 FET's in the U440 is important. So if red and green are the 2 FETs in one U440, then I would expect a big DC imbalance. (Maybe more than one can trim with VR101, Input Balance)

I will crack open my HM103, and do the requested measurements. I have to make some room on my bench .
(This HM103 was a gift to a Hameg collecting friend. But the friend died, and now this HM103 is back with me)

regards, Gertjan.

[Greybeard]

So are the red and green measurements of one FET in two different U440's? Or are red and green the 2 FETs in one U440?

These are two different U440s.
The pair in one U440 matches perfectly.
The pair in the other U440 also matches perfectly (but is shifted to the first U440).
Both U440s are in spec (see edit of my last post).

I tried to go your way and installed two (not particularly well-matched) BF245 (A instead of B) from my spare parts box into the device again.
The oscilloscope works, but correct symmetry adjustment is not possible, causing the line to move significantly vertically when changing the Y deflection factor.
I'll try to find better matching BF245s in my spare parts box.
Unfortunately the BF245 and its SMD successor, the BF545, have been discontinued, just like the U440.

By the way:
What do you think is the bandwidth bottleneck of the HM103 (10 MHz) if the preamplifier is capable of 60 MHz?

[Gertjan]

So are the red and green measurements of one FET in two different U440's? Or are red and green the 2 FETs in one U440?

These are two different U440s.
The pair in one U440 matches perfectly.
The pair in the other U440 also matches perfectly (but is shifted to the first U440).
Both U440s are in spec (see edit of my last post.

OK, all clear.

First thing I would check that nothing in the preamp has drifted out of spec. Also check for bad/loose solder joints. (my HM103 had many). One would assume that the circuit was designed to work with any U440, so maybe a resistor has drifted?
Then I would insert the U440 that works correct, and measure the voltage over R114 and R115. (Effectively measuring the current through the FET's). These should be equal.

Then insert the other U440, and measure U over R114 and 115 again. You should see the problem / difference... Adjust R114 and R115 for the same current as with the "good" U440.

From my notes: I measured -5,2V on top of both R114 and R115. (so 6,8V across = 2,48mA). Not sure though if this was already with the BF245B's...

Could you please do me a favor and measure the voltages at pins 1 and 14 of the uA733 (Y input = GND)?

Both pins are at -0,5VDC. No signal, and with any position of the input attenuator.

By the way:
What do you think is the bandwidth bottleneck of the HM103 (10 MHz) if the preamplifier is capable of 60 MHz?

I just measured the bandwidth from input to display: 15MHz -3B, with a test signal of 1Vtt.
So probably the Y-final amplifier ( or the tube itself) is the bottle neck.
But with a time base with max 0,5uS/sec, anything above 1MHz is getting cramped anyway...

The kicker: When I powered up my HM103 it was broken again .
The problem seemed to be in the Y-final amp. After taking some measurements, I looked at the base of the tube. and there I saw a wire of an Y plate broken loose....
So, when time permits I will add a small new chapter to the HM103 repair topic on Circuits Online .
And it underlines that bad solder joints are a common problem in my HM103...


regards, Gertjan.

[Greybeard]

Could you please do me a favor and measure the voltages at pins 1 and 14 of the uA733 (Y input = GND)?

Both pins are at -0,5VDC. No signal, and with any position of the input attenuator.

Thank you!
My values with the working U440:
+0.76V each.
Not very well-centered in the U440 specified input range of +/-1.0V, but I think the headroom of 0.24V / 5mV/div = 48 divs of possible deflection (while the screen is only 8 divs high) is enough.
So I will not try to correct.

First thing I would check that nothing in the preamp has drifted out of spec. Also check for bad/loose solder joints. (my HM103 had many).

I've still checked the attenuator and all connections of the Y-preamp, because of your excellent article "from hell". 
Everything seems to be OK.

One would assume that the circuit was designed to work with any U440...

That's of course is exactly what I thought, but... (see below)

...and measure the voltage over R114 and R115. (Effectively measuring the current through the FET's). These should be equal.
From my notes: I measured -5,2V on top of both R114 and R115. (so 6,8V across = 2,48mA). Not sure though if this was already with the BF245B's..

The values ​​are the same on my HM103: -5.2 V with the working U440.

Then insert the other U440, and measure U over R114 and 115 again. You should see the problem / difference... Adjust R114 and R115 for the same current as with the "good" U440.

OK, but in the meantime I foung another solution (see below).

By the way:
What do you think is the bandwidth bottleneck of the HM103 (10 MHz) if the preamplifier is capable of 60 MHz?

I just measured the bandwidth from input to display: 15MHz -3B, with a test signal of 1Vtt.
So probably the Y-final amplifier ( or the tube itself) is the bottle neck.
But with a time base with max 0,5uS/sec, anything above 1MHz is getting cramped anyway...

That's correct...

I will crack open my HM103...

You should not crack it literally... 

My solution:

I looked at the schematic of the donor device HM203-4:



The JFET input stage is build like a voltage devider with output at the center point, one JFET wired to +12V over two 332 Ohm resistors and the other JFET wired to -12V over the two other 332 Ohm resistors.
So, if one JFET drifts in one direction, the drift of the thermically coupled second JFET moves in the same direction, so the drift is compensated.
That compensation is also working for series production variation of the U440.
The pre-amp circuit is much simpler compared to the HM103 (no CA3046 needed at all) and works better!

I installed my "defective" U440 into the HM203, and after a small symmetry adjustment, the HM203 worked perfectly again.
I've measured the DC input voltages (pin 1+14) at the uA733: just a few millivolts for both channels.

So my solution is to simply swap the U440s of the HM103 and the HM203.

Nice and simple... 

[Gertjan]

So my solution is to simply swap the U440s of the HM103 and the HM203.

Nice and simple... 

Hi Greybeard,

Seems like a perfect solution to me .
Maybe Hameg did also some U440 selection...

The HM203-4 input is a classic circuit: the top FET is a source follower, with the bottom FET being a constant current sink, acting as active load on its source. The active load reduces non-linearities. When using a duo-FET,  the current sink FET tracks the source follower with temperature changes, so the offset remains near zero  independent of temperature.

I do not really understand why Hameg elaborated this circuit for the HM103....
Now the current sinks are transistors in a separate housing, so no auto temperature compensation.
The bottom FET is now used for setting the input DC balance. Maybe it was more important to have that temperature-track with the input follower?
(The two current sink transistors are matched and in in common housing, so they temperature-track each other too...)

In any case, Hameg used a different design for the HM103-2. But with the HM103-3 they returned to the original HM103-1 design, so it should have advantages .

regards, Gertjan.

[Greybeard]

I did a quick test with my HM103-3 yesterday.
I'm a little disappointed that it only has a -3 dB bandwith of 8.2 MHz.
At first, I thought it was caused by the function generator (it delivers a maximum of 15 MHz) and checked it with a 60 MHz HM605, but the generator was fine.

--------------------------------------------------------------------------------------------------------------------------------------------------------------

The first HM103 version from 1982 did not have any externally visible version indicators like -1, -2, or -3.
It had a hexagonal 6x7 divs CRT screen (1div=1cm) and large black AC/DC/GD push-buttons and was produced at least until mid-1984.


HM103-1

Inside, on the CRT shield, there was an orange sticker labeled "1031" (what means HM103-1).
On later HM103-1 models, the front panel plastic frame changed from dark gray color to light brown, like on the HM103-2.
I'm still looking for a manual with a circuit diagram for this unit (paper or PDF); I only have a photo of the first manual page with the print date 10/82.

--------------------------------------------------------------------------------------------------------------------------------------------------------------

The HM103-2 probably appeared in mid 1984 (after the last HM103-1 with AC/DC/GD push-button switch from my picture collection), but perhaps not until 1986 (the last changes to the manual; I've only seen index -2 models from 1986 so far).
It still has the 6x7 CRT screen and, as far as I know, is identical to the HM103-1 except for the new AC/DC/GD slide switch and the new input switch PCB.


HM103-2

It still had the original HM103 marking on the front panel, and initially, it had no index marking on the outside, but an orange sticker on the CRT shield with the designation "1032" (what means HM103-2).
The later HM103-2 models, however, already had a silver serial number sticker on the back (and inside) also with date of manufacture and the full model number HM103-2.
The accompanying manual already shows the new AC/DC/GD slide switch in the image on page 1 and in the Y-Input schematic, but the "old" front view of the HM103-1 with the AC/DC/GD push-buttons is still shown in the center of the manual.
The manual has a printing date of 12.83, but the circuit diagram is datet from 2.86 (so it must have been updated again, meaning the unit was also produced until at least 2.1986).
The schematics still has the designation HM103-1, as it was presumably adopted unchanged from the first HM103-1 (except for the change in the input slide switches).

--------------------------------------------------------------------------------------------------------------------------------------------------------------

The HM103-3 was released in late 1986 or early 1987, recognizable by the silver serial number sticker on the back.
It still had the HM103 marking without index on the front panel.


HM103-3

The HM103-3 is essentially identical to the HM103-2.
The only changes were the replacement of the 6x7 CRT and the corresponding front frame with a new rectangular 8x10 CRT.

The corresponding manual (printing date 1.1986) is titled HM103 without index and is identical to that of the HM103-2, except photos and CRT type name.
The schematics (printing date 5.1989) still has the name HM103-1, but the CRT has been changed to 8x10 (ER100, ER101 in the diagram; in fact, 95FB31F and ER102 were also available).

--------------------------------------------------------------------------------------------------------------------------------------------------------------

The HM103-2A was released in 1991.
It still features an 8x10 CRT, a new input circuit (only 10 Y attenuation steps), but also an additional Y-Mag./x5 switch and smaller gray AC/DC/GD push-buttons instead of the slide switch.
The calibrated position of the Y-Var control is now fully clockwise instead of the previously counterclockwise position.


HM103-2A

The new model name HM103-2 (without A) is printed on the front panel, but the index A can be found on the serial number sticker on the back and inside.
The manual and circuit diagrams also refer to HM103-2 without the index A and was printed in 4.1991.
The input amplifier has been completely redesigned and no longer includes a CA3086, but a TL051.
I am not aware of any device variant of this type without an A index on the serial number sticker.

It remains a mystery why Hameg assigned the index -2A and not simply -4.
Technically speaking, the HM103-2A is the version is farest away from the original HM103-2.

Very strange…

--------------------------------------------------------------------------------------------------------------------------------------------------------------

All of this wasn't easy to figure out, as there were some units that were incorrectly labeled by Hameg (either old labels stuck to already new units, or labels for the new series stuck to the last units of the old series—I've seen both).

Please don't take all this information as 100% accurate, if someone has better information, please let me know.

EDIT: updated history data.

[Gertjan]

I did a quick test with my HM103-3 yesterday.
I'm a little disappointed that it only has a -3 dB bandwidth of 8.2 MHz.

Hi Greybeard,

Indeed, 8.2MHz bandwith is a bit dissapointing. And below spec.
Could it be that the bandwith is limited because of over-compensation? For example by VC300 in the Y final amplifier?
The bandwith I measured on mine is after I trimmed the scope for as perfect as possible square waves.

You could measure the bandwith on the output of the pre-amp (also for checking the compensation on the input attenuator). And then see if you can optimise things with VC300.
(In my case I had to remove a parallel capacitor on the back of the PCB)


Thank you for the interesting info on HM103 history. I did not know.

It is a thing that makes repairing interesting: You dive into the history and schematics, learning all kinds of things along the way .

I posted an update about my HM103-3 repair in my "Hameg HM103-3 repair from hell" topic. Also referring to your topic with links.


regards, Gertjan.

[rob040]

Did you know that in 1982 there was an HM103 without the -1, -2, and -3 index? It had a 6x7 screen diagonal (hexagonal) and large black AC/DC buttons?
In 1983, the HM103-1 came out. It still has the 6x7 screen. The manual still shows the HM103 on the first page, but the front view in the middle shows the new AC/DC slide switch.
In 1987 1991, the HM103-2 came out (manual dated 4.91). It features the new 8x10 rectangular CRT, new input circuitry (only 10 Y steps), but also an additional Y-Mag./x5 switch and small gray AC/DC pushbuttons instead of the slide switch.
In 1990, the HM103-3 arrived. It's the same unit as the HM103-1. The only changes were the 6x7 CRT and the front frame were replaced by the 8x10 CRT and front frame from the HM103-2.
In 1987 1991, the HM103-2 came out (manual dated 4.91). It features the new 8x10 rectangular CRT, new input circuitry (only 10 Y steps), but also an additional Y-Mag./x5 switch and small gray AC/DC pushbuttons instead of the slide switch.

All of this wasn't easy to figure out, as there were some units that were incorrectly labeled by Hameg (either old labels stuck to already new units, or labels for the new series stuck to the last units of the old series—I've seen both).

Funny, I did a similar exercise a while ago and made an overview to have more inside in the way Hameg was numbering this oscilloscope:

So, after -3 they came with the -2A. I didn't find a HM103-1, but I'm not surprised they exist.

Did you know that there is also a Hameg HM103 tube-voltmeter? They probably forgot about that one when introducing the HM103 oscilloscope. 

[Greybeard]

I compared your list with my data. Obviously you used the same data source.
Have you ever seen a HM103-2 backlabel without the A?
Do you have any pictures of it (->PM?)?

I didn't find a HM103-1, but I'm not surprised they exist.

The HM103-1 had no external -1 index. It is inside (probably on a yellow-colored sticker on the CRT shield).

Did you know that there is also a Hameg HM103 tube-voltmeter? They probably forgot about that one when introducing the HM103 oscilloscope. 

Yes I know.
There were also HM203 oscilloscopes and HM203 volt-ohmmeters.
Did they run out of numbers? I was asking myself what they were smoking.   


PS: I've reworked my history above a bit.

[rob040]

Have you ever seen a HM103-2 backlabel without the A?

Yes, I own one, see the following pictures:

[Greybeard]

Thanks, Rob.
In the meantime I've found a few other HM103-2 from 1986 (like yours), but none from before 1986.

The HM103 manual (dated 12.83) shows the version with a 6x7 screen and AC/DC/GD slide switch (=HM103-2) at the beginning of the manual, but in the middle it still shows the old front view of the HM103-1 with the AC/DC/GD push button switches.
The schematics in this manual are labeled HM103-1 and dated 12.83 - except the section with the slide switch and the page with the trigger/timebase circuit (dated 2.86), so probably there was an update for the HM103-2 in 1986.
Interestingly, the 1986 pages still bear the HM103-1 marking... 

Now I'm wondering when the first ones with slide switches (HM103-2) actually appeared and whether there were any before 1986.
The last HM103-1 ("1031" label inside, push button switch) I could find was from 06/84, SN 94xx.

Are you sure it was "1032" and SN 10xx?