Rohde&Schwarz URV3 Millivoltmeter

Internal
Page content

High Frequency Millivoltmeter for range 10KHz..2GHz. It came with a probe for range 1KHz..1600MHz. Sold as defective.

EBay Images

Vendor id is 302.9014.02 . Device reacts somehow on input, but does not display anything meaningful.

s-l1600 (3).webp

s-l1600.webp

Meßkopf: HF-Durchgangskopf 1KHz-1600MHz, 50 Ohm: 243.9418.55 FNr. 870271/39

Dämpfungsglied: 0-2400MHz 10dB, 50 Ohm, 3Watt.

RBD BN 33661/50 Fnr. F2690/6

s-l1600 (4).webp

Repair story (not finished)

Fixing power supply section

The Device was sold as defective. On the first inspection, I found that there was some extremely bad “repair” in the past. One transistor was obviously replaced. The soldering quality of the repair job looked very poor.

I started with checking power supply. This device has 4x1,5 volts batteries, and creates from those 6 volts in a first step +/-26 volts unregulated, and from that +/-15 volts regulated. While +15 volts are ok (measured 14.8 volts), the -15-volts rail was wrong, with about -26 volts.

The transistor replaced by someone else was part of the -15 volts rail. It regulates the -15 volts based on the +15 volts rail stable output. I found that for the main regulating transistor on the defective rail, input is about 26.3 volts and output slightly below, so 25.6 volts.

Examining the poor soldering, I found that three PCB traces were lost due to the poor unsoldering/soldering when replacing the transistor. Obviously, the repair was executed without unsoldering tools and low soldering ability.

The biggest surprise was next, that the transistor replacement was a totally wrong type. Required was BCY79 (PNP), but someone inserted BCY59 (NPN). This could never have worked. So I removed the wrong transistor, inserted some BC557 for testing and fixed the missing PCB copper traces by adding small wire connections. After that, I had also -14.8 volts on the negative rail.

The strange “repair” from the past also included replacement of a 47µF capacitor, done with 2x22µF capacitors, soldered together to create a 44µF cap. Also soldered very ugly. I replaced that with a real 47µF capacitor.

Finally, there was some capacitor (470µF/16 volts) with some bulge, in section where unregulated voltages are created, which was also replaced, despite its electrical values were still looking fine.

After these PSU fixes, the device still did not improve regarding measuring AC. Hm, there is more to check.

Checking more sections

To go deeper into error analysis, we need to understand how the URV3 device works.

Theory of operation

Below is a block-style schematic of an older URV meter. I could not find such a schematic for the URV3. But I guess the difference, at the block level, is small.

urv-blockschaltbild.png

The URV3 compares two DC voltages. One is the rectified voltage to be measured, the other voltage is generated in AC in the device and also rectified. Both rectifications are done inside the measurement probe (“Meßkopf”), and the two rectifiers are selected to behave as nearly identical. Inside the measurement probe, the difference of these two voltages is generated (see the “+” sign in the block-style schematic). The difference can be tiny (less than 1µV). This difference is being amplified by a magnitude of about $$10^{6}$$ in the device.

The chopper amplifier magic

Because a simple amplifier usually has some DC drift, let’s say in the range of some ten µV, it is not feasible to use a DC amplifier here. The signal we want to measure will simply disappear in the intrinsic drift of the amplifier. Nowadays technology offers super low-drift OpAmps that could be used here, but at the time the URV3 was designed (1970ies), these chips were not available.

Side note: Another device using the chopper approach is the venerable HP3400A. For this device, manufactured through decades, HP started with a chopper amp solution, but later, when low-drift OpAmps were approaching the market, they moved to a more modern approach. I have looked at that remarkable device and replaced some old chopper PCB with a newer one here

So there was the task to amplify the DC voltage by magnitudes like $$10^{6}$$, but to keep the intrinsic amplifier DC drift out of the measurement.

A common solution to amplify tiny DC voltages is to convert the DC voltage into an AC voltage, which can easily be amplified, and DC drift is not an issue there. At a later stage, let’s say when the voltage has been amplified up to about 1 Volt, we can convert the AC voltage back to DC.

A common approach to converting a DC voltage into an AC voltage is to use a chopper circuit. The chopper amplifier chops the DC voltage into an AC voltage, by applying some chopping frequency (AC) onto it. Let’s say we have some DC voltage, fixed to 20µV. The chopper runs with a frequency of 25 Hertz and shortens the DC signal 25 times per second. This will result in a rectangle waveform, jumping between 0µV and 20µV. If we add a DC offset of -10µV, we get a nice AC signal that has peaks at -10µV and +10µV.

In the block-style schematic, the chopper is shown as a switch signal in a box. In signal path, the first switch box creates the AC voltage from the DC voltage. The AC amplifier follows. By applying the same chopper frequency to the next switch box, the DC signal is regenerated, from a waveform perspective. But now, the signal is largely amplified. And all drift issues were eliminated.

What follows is some additional amplification to drive the meter and the DC output. Because we have “large” DC voltages, let’s say in range 10mV .. 1V, the drift voltages cannot harm in these later stages.

Feedback path

So far we have looked at the amplification path of the device. The whole device is designed as a feedback loop. This means we have at the lower part of the schematic all blocks that create the feedback loop.

The loop tries to generate an AC voltage that is nearly equal to the signal measured. In fact, it generates an AC voltage that would be shown as zero value at the meter, depending on range selected. The difference is then displayed at the meter as the voltage measured.

To create some AC voltage, a 5KHz signal source is used, creating some rectangle signal. According to the selected range, its duty cycle is defined (this is the box with the fixed resistor inside). The waveform is created using another chopper element.

From that rectangle waveform, a sinusoid waveform is generated (band pass box). With some amplification and through some transformer, the AC voltage is routed to the measurement probe, where it is finally rectified.

Manual:

Other docs: