Signal ground decoupled from chassis or not? - UK Vintage Radio Repair and Restoration Discussion Forum

Uncle Bulgaria · 23rd Aug 2020, 12:09 am

Another Uher question. I thought I'd fixed up my Miniline stack, and I've been listening to it a lot.

Today I discovered a buzz and hum, and realised it disappeared when I lifted the preamp unit away from the power amp. The two are connected with two DIN cables - one for power and relay signals (on/off and 'speaker relays are in the power amp, controlled from the pre-amp) and the other for audio. Measuring the ground of these DIN cables to the chassis of both amps gives readings of a couple of Ohms.

Switching on, taking the same readings shows 20 Ohms to the power amp chassis, and 0.9Meg to the preamp chassis. No wonder there was a problem when they came into contact!

Either they are supposed to be like this, or there is a fault. My guess, without investigation as yet, is that there's a failed decoupling capacitor in the power amp. Is this reasonable? It seems unlikely that Uher would rely on a couple of millimetres of rubber stick-on foot to isolate the pre-amp from the power amp, when the latter has heatsinks that stick up and catch the pre so are likely to come into contact with the chassis.

Silicon · 23rd Aug 2020, 9:51 am

That difference in resistance between power off and power on is puzzling.

Do you have a method of measuring the resistance that is immune to AC and DC voltages across the resistance?

Uncle Bulgaria · 23rd Aug 2020, 2:46 pm

I'm not sure what you mean by that. I'm measuring with a DMM between the ground wire at the Z140 signal end, to chassis screws on the front panel of the VG 840 and Z140. With only the 6-pin power cable connected between the two (to actuate the relay from the VG 840 switch), the result is the same.

The resistance between the ground of the signal cable and the central ground pin of the power interconnect is approx. 5k.

Between the power ground of the DIN socket and the VG 840 chassis (not connected to the Z140) it climbs to ~25k. The same measurement to the Z140 chassis sinks to 7k.

It seems like current is flowing somewhere, because if I allow the chassis to touch, measuring sub-Ohm on the VG840, if I quickly remove and replace the probe having separated the chassis then the resistance quickly climbs each time until it reaches ~1M. For the duration of the probe's application the resistance doesn't change.

I'm loath to strip down the Z140 again until I know what I'm looking for as it's quite tightly packed in!

Uncle Bulgaria · 15th Sep 2020, 2:56 am

I'd like to bump this again to see if anyone has any further thoughts. Does the idea of a decoupling failure, creating a DC mismatch between the ground potentials on the two boxes, hold water? When I get my capacitor tester working, should I be looking at that? The circuit diagrams for both units are attached. I'm sorry they're blurry - it's the forum limits, but I hope my theory can be considered worthless or worthwhile from topology rather than detailed examination of values!

Diabolical Artificer · 15th Sep 2020, 7:00 am

To clarify, you've got a stack of hifi, amp, preamp and so on, with the amp sitting on top of (BTW does anyone remember the monkey Topof on Hartley the Hare? Sorry, I digress) the preamp you have a hum/buzz, therefore the chassis of each which is mains earthed referenced and thus maybe ground referenced are completing a circuit, yes?

So, straight off, most hifi or test gear have feet to they are isolated from it's neighbor, is this not the case here? If they are touching and grounding to each other what does slipping a insulator between the two do?

Too complicate matters each unit is connected by a DIN cable which are grounded when the units are powered off, but when power is applied, the preamp shows a 900k resistance between DIN cable gnd and unit gnd, therefore you need to find out why this is, that's what Silicon was referring to.

Some buzzing out with a DMM on diode mode might be beneficial trying to find the break or area of high resistance.

Andy.

Uncle Bulgaria · 15th Sep 2020, 12:37 pm

Thank you for the thoughts. It's not quite a normal stack - the problem is between the Z140 and the VG840, which are only connected by the DIN cables. The Z140 is power amplifier and SMPS which is connected to the mains, and has a DIN output to power the VG 840 preamp, and a DIN input to take signal from the VG 840. See pic. The tuner (not connected) has a further DIN connection back to the VG 840, so is powered by a daisy chain back to the Z140. There is only the one mains connection, which made me think it's not the usual ground loop problem.

Yes, the VG840 has feet, but they are very small and the heatsinks of the Z140 stick up so the clearance is minimal. The problem arises when the chassis touch. It seems unlikely that if this were a crucial element of the design that it would be so marginal, especially as the ground references should surely be shared by the whole stack.

Diabolical Artificer · 17th Sep 2020, 5:41 am

Feet get lost, have they been changed in the past for smaller ones? It could also be a capacitive coupling problem, the two units chassis essentially are two big metal plates forming a cap. If it were me I'd be looking at the DIN lead grounds and go back to your original diagnosis, why does the DC res go up on power on? The 20r and 900k could be forming a voltage divider or the 900k is developing an AC voltage current across it, hence hum, just a wild guess, needs investigation.

Andy.

Uncle Bulgaria · 24th Nov 2020, 4:35 pm

I've had both apart again and there was a high-resistance break in the ground plane between the front of the VG840 and the rear plate. The ground passes through one of the on-board connectors, which I cleaned rigorously. Now the problem remains but is greatly lessened. I measure 13mV chassis-chassis when they are not touching, and there is no buzz in the headphones. This decreases, as imagined, when they touch to 0.5mV, but the buzz is immediately present. It doesn't blow your head off as it used to, though. The oscilloscope shows some buzz on the ground plane, probably from the SMPS.

I wonder if it is a capacitive coupling problem. I've tested the ground continuity on both units and removed and tested about all the coupling capacitors and possible routes for signal and SMPS noise to join up, with no conclusive results.

23rd Aug 2020, 12:09 am	#1
Uncle Bulgaria Nonode Join Date: May 2016 Location: Cornwall, UK. Posts: 2,333	Signal ground decoupled from chassis or not? Another Uher question. I thought I'd fixed up my Miniline stack, and I've been listening to it a lot. Today I discovered a buzz and hum, and realised it disappeared when I lifted the preamp unit away from the power amp. The two are connected with two DIN cables - one for power and relay signals (on/off and 'speaker relays are in the power amp, controlled from the pre-amp) and the other for audio. Measuring the ground of these DIN cables to the chassis of both amps gives readings of a couple of Ohms. Switching on, taking the same readings shows 20 Ohms to the power amp chassis, and 0.9Meg to the preamp chassis. No wonder there was a problem when they came into contact! Either they are supposed to be like this, or there is a fault. My guess, without investigation as yet, is that there's a failed decoupling capacitor in the power amp. Is this reasonable? It seems unlikely that Uher would rely on a couple of millimetres of rubber stick-on foot to isolate the pre-amp from the power amp, when the latter has heatsinks that stick up and catch the pre so are likely to come into contact with the chassis.

23rd Aug 2020, 9:51 am	#2
Silicon Nonode Join Date: Jun 2007 Location: Coulsdon, London, UK. Posts: 2,163	Re: Signal ground decoupled from chassis or not? That difference in resistance between power off and power on is puzzling. Do you have a method of measuring the resistance that is immune to AC and DC voltages across the resistance?

23rd Aug 2020, 2:46 pm	#3
Uncle Bulgaria Nonode Join Date: May 2016 Location: Cornwall, UK. Posts: 2,333	Re: Signal ground decoupled from chassis or not? I'm not sure what you mean by that. I'm measuring with a DMM between the ground wire at the Z140 signal end, to chassis screws on the front panel of the VG 840 and Z140. With only the 6-pin power cable connected between the two (to actuate the relay from the VG 840 switch), the result is the same. The resistance between the ground of the signal cable and the central ground pin of the power interconnect is approx. 5k. Between the power ground of the DIN socket and the VG 840 chassis (not connected to the Z140) it climbs to ~25k. The same measurement to the Z140 chassis sinks to 7k. It seems like current is flowing somewhere, because if I allow the chassis to touch, measuring sub-Ohm on the VG840, if I quickly remove and replace the probe having separated the chassis then the resistance quickly climbs each time until it reaches ~1M. For the duration of the probe's application the resistance doesn't change. I'm loath to strip down the Z140 again until I know what I'm looking for as it's quite tightly packed in! Attached Thumbnails

15th Sep 2020, 7:00 am	#5
Diabolical Artificer Dekatron Join Date: Jul 2009 Location: Sleaford, Lincs. UK. Posts: 7,658	Re: Signal ground decoupled from chassis or not? To clarify, you've got a stack of hifi, amp, preamp and so on, with the amp sitting on top of (BTW does anyone remember the monkey Topof on Hartley the Hare? Sorry, I digress) the preamp you have a hum/buzz, therefore the chassis of each which is mains earthed referenced and thus maybe ground referenced are completing a circuit, yes? So, straight off, most hifi or test gear have feet to they are isolated from it's neighbor, is this not the case here? If they are touching and grounding to each other what does slipping a insulator between the two do? Too complicate matters each unit is connected by a DIN cable which are grounded when the units are powered off, but when power is applied, the preamp shows a 900k resistance between DIN cable gnd and unit gnd, therefore you need to find out why this is, that's what Silicon was referring to. Some buzzing out with a DMM on diode mode might be beneficial trying to find the break or area of high resistance. Andy. __________________ Curiosity hasn't killed this cat...so far.

15th Sep 2020, 12:37 pm	#6
Uncle Bulgaria Nonode Join Date: May 2016 Location: Cornwall, UK. Posts: 2,333	Re: Signal ground decoupled from chassis or not? Thank you for the thoughts. It's not quite a normal stack - the problem is between the Z140 and the VG840, which are only connected by the DIN cables. The Z140 is power amplifier and SMPS which is connected to the mains, and has a DIN output to power the VG 840 preamp, and a DIN input to take signal from the VG 840. See pic. The tuner (not connected) has a further DIN connection back to the VG 840, so is powered by a daisy chain back to the Z140. There is only the one mains connection, which made me think it's not the usual ground loop problem. Yes, the VG840 has feet, but they are very small and the heatsinks of the Z140 stick up so the clearance is minimal. The problem arises when the chassis touch. It seems unlikely that if this were a crucial element of the design that it would be so marginal, especially as the ground references should surely be shared by the whole stack. Attached Thumbnails

17th Sep 2020, 5:41 am	#7
Diabolical Artificer Dekatron Join Date: Jul 2009 Location: Sleaford, Lincs. UK. Posts: 7,658	Re: Signal ground decoupled from chassis or not? Feet get lost, have they been changed in the past for smaller ones? It could also be a capacitive coupling problem, the two units chassis essentially are two big metal plates forming a cap. If it were me I'd be looking at the DIN lead grounds and go back to your original diagnosis, why does the DC res go up on power on? The 20r and 900k could be forming a voltage divider or the 900k is developing an AC voltage current across it, hence hum, just a wild guess, needs investigation. Andy. __________________ Curiosity hasn't killed this cat...so far.

24th Nov 2020, 4:35 pm	#8
Uncle Bulgaria Nonode Join Date: May 2016 Location: Cornwall, UK. Posts: 2,333	Re: Signal ground decoupled from chassis or not? I've had both apart again and there was a high-resistance break in the ground plane between the front of the VG840 and the rear plate. The ground passes through one of the on-board connectors, which I cleaned rigorously. Now the problem remains but is greatly lessened. I measure 13mV chassis-chassis when they are not touching, and there is no buzz in the headphones. This decreases, as imagined, when they touch to 0.5mV, but the buzz is immediately present. It doesn't blow your head off as it used to, though. The oscilloscope shows some buzz on the ground plane, probably from the SMPS. I wonder if it is a capacitive coupling problem. I've tested the ground continuity on both units and removed and tested about all the coupling capacitors and possible routes for signal and SMPS noise to join up, with no conclusive results.