100:1 probe for measuring ripple in a tube amp power supply

[Andreax1985]

I have successfully measured ripple using an external capacitor and DC coupling. If you are interested in seeing the results and following my "investigation", I started a new thread in the Beginners section:

https://www.eevblog.com/forum/beginners/tracking-back-100hz-hum-in-a-tube-guitar-amp/

[Peter S]

Hi Folks,  Just joined this forum specifically for this thread.   That is exactly my dillema;  working on tube amps with 460 V (and  higher)  B+,  looking for ripple.   My scope is also 400V max DC+ AC.   Some very interesting points learned here.  If I place a probe on a 460V  B+,  there will be a momentary 460V spike that would be passed by a DC blocking capacitor.
   At the risk of beating a dead horse,  why not just build a simple resistor voltage divider,  two  1 Meg resistors with the scope at the centre tap.   Would this not reduce everything to safe limits,  AC and DC?
   Thanks,  Peter in Canada.  Sorry if this is redundant

[bdunham7]

Hi Folks,  Just joined this forum specifically for this thread.   That is exactly my dillema;  working on tube amps with 460 V (and  higher)  B+,  looking for ripple.   My scope is also 400V max DC+ AC.   Some very interesting points learned here.  If I place a probe on a 460V  B+,  there will be a momentary 460V spike that would be passed by a DC blocking capacitor.
   At the risk of beating a dead horse,  why not just build a simple resistor voltage divider,  two  1 Meg resistors with the scope at the centre tap.   Would this not reduce everything to safe limits,  AC and DC?

I've never seen this thread before, but it is worth commenting on.

First, the original OP had a Siglent SDS1104X-E and thus the whole worry about the DC blocking capacitor was irrelevant--that scope, along with many modern DSOs, has a 1M DC-coupled input impedance regardless of whether DC or AC coupling is selected.  So he could have simply used the standard 10X probes and AC coupling and taken his reading.

Second, if you do have a scope with true DC blocking capacitor that gets charged up to the DC bias present on the input, the safest solution is instead of a 100X/100M probe that is basically a 99M resistor, use a 100X/10M probe that has a built-in voltage divider that will result in the DC bias at the scope of ~0.11X whatever is at the probe tipe.  The Probemaster 4910-2 works this way, as do a number of older Tek probes.

Of course this still leaves you with 100X attenuation, which may be too much.  In that case, if you can live with a very limited BW, you can make your own divider probe.  I'd suggest something like a 4.5M and 1M resistor in series with the scope across the 1M, that gives you 5M total input resistance and 10X voltage division with a 5.5X reduction of the DC component across the scope's capacitor.

[magic]

   At the risk of beating a dead horse,  why not just build a simple resistor voltage divider,  two  1 Meg resistors with the scope at the centre tap.   Would this not reduce everything to safe limits,  AC and DC?

You could, but the scope likely contains another 1MΩ to ground internally so division ratio would be 3:1 rather than 2:1.
Frequency response would be limited due to input capacitance of the scope, probably good enough for 100Hz or similar.
Real passive probes contain additional (partly adjustable) capacitors for tuning AC response to match DC response, that's the whole "probe compensation" thing.

[Kleinstein]

The Scope would only have 1 M to ground. So a 1 M resistor would give a 1:2 divider, not a 1:3 divider.
The Question is a little if the AC coupling before / after the resistors. Old style there was just a capacitors beform. Newer could split the 1 M and have something like 10 M to ground before the capacitor and 1.1 M after the capacitor. So one would still discharge the capacitor and still have not less than 1 M impedance to ground.

Jumping between 0 V and 460 V with AC coupling could still cause significant transient spikes. Many scopes will survive, but I have damages a scope input this way (go from some +150 to -150 V on a scope with specs for 200 V max.)
If possible use a 10:1 probe to be on the safe side, though the AC coupling capacitor could still charge up to the full DC votlage !

For ripple voltage measurements one sometimes looks for pretty low AC voltage on top of high DC. So a 1:100 divider is not ideal.

[Fungus]

At the risk of beating a dead horse,  why not just build a simple resistor voltage divider,  two  1 Meg resistors with the scope at the centre tap.   Would this not reduce everything to safe limits,  AC and DC?

The impedance of the average oscilloscope input is also about 1Meg so when you connect up your probe you'll be loading the circuit down.

A 100:1 probe will have a very high impedance (99MOhms) so that won't happen.

You can do it with resistors but you have to take all the impedances into account.

Make sure your probe is rated for high voltage. There are 100:1 probes out there which are sold for their impedance properties, not their high voltage abilities.

[TimFox]

Again, I hate to re-animate a dead horse, but I have successfully used a good polypropylene capacitor and normal metal-film resistor to form a high-pass network to pass frequencies above, say, 10 Hz to measure ripple on a Vbb supply at a few hundred volts.

For AC at 50 to 120 Hz, this is essentially 1:1, and the resistor can be, say 100 kohm to get only a 10% fudge factor (frequency response only) with a scope impedance of 1 megohm.
For the high-pass capacitor, 220 nF or so will suffice.
The AC loading of 100 kohms should be negligible for a Vbb rectifier circuit.

As always, proper high-voltage safety precautions must be taken.

[tautech]

Again, I hate to re-animate a dead horse, but I have successfully used a good polypropylene capacitor and normal metal-film resistor to form a high-pass network to pass frequencies above, say, 10 Hz to measure ripple on a Vbb supply at a few hundred volts.

For AC at 50 to 120 Hz, this is essentially 1:1, and the resistor can be, say 100 kohm to get only a 10% fudge factor (frequency response only) with a scope impedance of 1 megohm.
For the high-pass capacitor, 220 nF or so will suffice.
The AC loading of 100 kohms should be negligible for a Vbb rectifier circuit.

As always, proper high-voltage safety precautions must be taken.

100% however for the novice I consider this a very risky practice as there is zero indication on the scope display or within measurements to remind the user they are working with elevated voltages.
Therefore best advice is to invest in 100x probes that will pass DC for the scope to clearly display the elevated voltages worked with.
As a secondary reminder but primary to the user is selection of a brightly colored probe as it's not just about getting the job done but protecting yourself and the instrument.
Recommend these if you can find them locally:
http://www.pintek.com.tw/productDetail/land-ctop-2/index/pscsn/17073/psn/19272

[TimFox]

Again, I hate to re-animate a dead horse, but I have successfully used a good polypropylene capacitor and normal metal-film resistor to form a high-pass network to pass frequencies above, say, 10 Hz to measure ripple on a Vbb supply at a few hundred volts.

For AC at 50 to 120 Hz, this is essentially 1:1, and the resistor can be, say 100 kohm to get only a 10% fudge factor (frequency response only) with a scope impedance of 1 megohm.
For the high-pass capacitor, 220 nF or so will suffice.
The AC loading of 100 kohms should be negligible for a Vbb rectifier circuit.

As always, proper high-voltage safety precautions must be taken.

100% however for the novice I consider this a very risky practice as there is zero indication on the scope display or within measurements to remind the user they are working with elevated voltages.
Therefore best advice is to invest in 100x probes that will pass DC for the scope to clearly display the elevated voltages worked with.
As a secondary reminder but primary to the user is selection of a brightly colored probe as it's not just about getting the job done but protecting yourself and the instrument.
Recommend these if you can find them locally:
http://www.pintek.com.tw/productDetail/land-ctop-2/index/pscsn/17073/psn/19272

In my projects building vacuum-tube preamps, I needed to check for ripple on the Vbb supply, but a 100:1 probe into a 5 mV/div scope input would not be sufficiently sensitive.
I did not build these hp filters as "probes", but as a circuit that could be connected to the power supply and CRO before turning on the supply.
Safety first!

[trobbins]

A well made or restored valve amp should have a bleed resistor across the B+ supply. Add a fixed divider resistor to that bleed, and confirm it provides a 10:1 or 100:1 ratio, and then just probe the safe end of the bleed. Make the probing process as safe as you can. I use that form of divider to extend to a maintence socket that can easily connect to an external meter for checking rails and bias levels.

[Peter S]

 Being close to Christmas,  I thought I deserved a new toy.     Personally,  I am leaning towards a simple voltage divider (maybe 2:1)  taking into consideration the internal impedance of the scope,  hopefully bringing things into safe limits.  To be even safer,  I thought I would buy myself a second scope to be my 'daily driver'  and save my Leader LBO-516 for 'special occasions'.  It is a 100 MHz scope with 8 trace capability and features mostly beyond me.  It has been working great for me for over 25 years.
   I found its poor cousin,  an LBO-522 for $160 CDN and snapped it up.  It has a max input voltage of 600V.   Most scopes I've seen  (like my 516,  are only 400V)    The only issue is the trace intensity has to be at or close to max for a good visible trace.   The manual shows internal adjustments for  intensity and HV.    Somehow I thought they would be closely related and the two pots  VR1  and VR2  are not far apart on the schematic.   The scope  otherwise seems to be perfectly calibrated.   I was hoping that a slight tweak of one of these controls would brighten up the trace....but probably more to it?   Just wondering if anyone had experience with this?   The CRT does not seem to have any dim areas that I would imagine would develop if it had been left on for years with a trace at the zero line.
     Thanks again everyone for the great information

[wasedadoc]

I would tack a 100nF cap (of adequate voltage rating) in series with a 4.7M resistor between the amp supply and zero Volts.  (Capacitor to supply and resistor to zero Volts.) After powering up amp wait 60 seconds before clipping an ordinary x1 scope probe across the resistor. Measure and note amplitude of ripple.  Disconnect probe before powering off amp.  Correct measured value to compensate slight reduction in measured ripple due to capacitor reactance at whichever of 50, 60, 100 or 120 Hz is appropriate.

[floobydust]

OP, I've measured ripple on tube amps with Leader scopes and I don't recommend it without a few precautions.
Once you select AC-coupling, the scope's input cap charges up to the DC potential and is a problem because there is no discharge path. So the either the cap holds a stray charge and kills the next thing probed (circuit under test or the scope's front-end if X1) or discharges (ka-bang!) into the AC/GND/DC switch, I think it was a LBO-510 (0.1uF) that does that.

What I've done is after the measurement, touch the scope probe to my hand, =a lazy man's 10MEG resistor to GND and the scope trace will go off screen for a few seconds and then come back to center. The scope's front-end is overloaded but at low current, so the input clamp is not overloaded. This is the cowboy way to discharge that cap and not recommended at all.
But it shows the problem - either that cap is dumped into the scope's front end or the next circuit element probed.

It's too bad this thread did not come up with a schematic for a simple add-on capacitor jig for such measurements.
I would not go 100:1 because you lose the ability to measure low level AC just because of the very large DC component. Even worse is 100:1 scope probes are cheap like borsch on Ali.

Does your Leader scope have the old Marcon oil-filled electrolytic caps in the HV section? These are known to croak with age and cause dim displays.
https://www.opweb.de/english/company/Leader/LBO-522

[coromonadalix]

Bk precision PR100A

Input resistance of 100Mohm (probe resistance 99 Mohm ±1%)
Input capacitance of 6.5pF
Maximum input voltage of 1200VDC including peak AC derating with frequency IEC cat I
Cable length of 1.2m


Bk precision  PR2000B

The PR2000B is a high voltage Oscilloscope Probe with 200MHz bandwidth.
2KV
100MΩ/5pF Input impedance

[jonpaul]

use AC..DC..GRD switch in GRD
Pos
attach 10x probe to tube
Turn on amp, allow B+ to precharge the scopes AC coupling cap
move switch to AC
enjoy

[wasedadoc]

It's too bad this thread did not come up with a schematic for a simple add-on capacitor jig for such measurements.
I would not go 100:1 because you lose the ability to measure low level AC just because of the very large DC component.

Read the post immediately above the one where you wrote that.  Did that really need a diagram in addition to the words?

[floobydust]

It's too bad this thread did not come up with a schematic for a simple add-on capacitor jig for such measurements.
I would not go 100:1 because you lose the ability to measure low level AC just because of the very large DC component.

Read the post immediately above the one where you wrote that.  Did that really need a diagram in addition to the words?

That works OK but remembering a four or five step procedure, with great calamity if not followed, may be too much.
It's common to measure ripple in many locations, or look at audio on the plate, in one session. I'm not turning power on/off every time.

use AC..DC..GRD switch in GRD
Pos
attach 10x probe to tube
Turn on amp, allow B+ to precharge the scopes AC coupling cap
move switch to AC
enjoy

You need the part where you discharge that AC coupling cap after disconnecting the probe.

[jonpaul]

We measure  PARD on HV DC or_B+ bus, with external cap,10x probe, cap discharge add 1m bleeder with a switch.   For RMS, use AC meter like Fluke, or P.

best référence on PS measurement is the 1980s Tektronix and  HP application note like AN90, power supply measurements.

Jon

[David Hess]

Not so old Tektronix oscilloscopes include a 1 megohm resistor in series with the coupling capacitor when the coupling switch is set to GND, which serves to precharge the coupling capacitor when switching between AC and DC coupling.

[Peter S]

Thanks again for all the info.  The online manual for the LBO-522 seems to be incomplete but the LBO-516 service manual  (hopefully related) had an adjustment procedure for the internal intensity control.  I followed this and all is great!   The pot required only 10 or 15 degrees of rotation to get brightness back to perfect.  There still seemed to be lots of range left on either side of the pot's current position so I will assume all is good.    Further;  if this scope has sat for years with the trace brightness limited by this control,  may it have extended the life of the CRT?  (just being optimistic here!).

   At the risk of wearing everyone's patience;   if I simply had voltage divider using two  1 Meg resistors with the scope attached to the centre-tap,  how is it possible for the DC potential at centre-tap (scope connection) to be anything more than half of the potential at the top (input) of the series string?
  I feel I may be 'booted' off this thread,  as  you have all tried so hard to explain this!    Merry Christmas and thanks again!
Peter

[floobydust]

Don't worry about asking questions here, we share what we know and can always learn more about the art 
My main concern is damaging vintage test equipment doing these measurements because there can be several traps which we are discussing.

You can add the 2:1 divider ahead of things, but it will actually read 1/3 because the scope ends up a 1MEG load in parallel with the 1MEG lower divider resistor giving 500k.
If you try measure signal on a gain stage's plate, it will get loaded down somewhat.

I realized the 100:1 probe is actually a bad idea. AC-coupled, a scope's input cap will charge to the full DC potential regardless of the probe attenuation, if it's 1x, 10x, 100x because the cap is ahead of the scope's 1MEG front-end resistor. This is a trap.


If you have any drama with the Leader, check it's PSU rail voltages are OK and then I go after the old Marcon caps.

[vk6zgo]

Thanks again for all the info.  The online manual for the LBO-522 seems to be incomplete but the LBO-516 service manual  (hopefully related) had an adjustment procedure for the internal intensity control.  I followed this and all is great!   The pot required only 10 or 15 degrees of rotation to get brightness back to perfect.  There still seemed to be lots of range left on either side of the pot's current position so I will assume all is good.    Further;  if this scope has sat for years with the trace brightness limited by this control,  may it have extended the life of the CRT?  (just being optimistic here!).

   At the risk of wearing everyone's patience;   if I simply had voltage divider using two  1 Meg resistors with the scope attached to the centre-tap,  how is it possible for the DC potential at centre-tap (scope connection) to be anything more than half of the potential at the top (input) of the series string?
  I feel I may be 'booted' off this thread,  as  you have all tried so hard to explain this!    Merry Christmas and thanks again!
Peter

Of course, it can't be ---it will always be less than half, as the 'scope input Z will be in parallel with the bottom resistor.

[vk6zgo]

Don't worry about asking questions here, we share what we know and can always learn more about the art 
My main concern is damaging vintage test equipment doing these measurements because there can be several traps which we are discussing.

You can add the 2:1 divider ahead of things, but it will actually read 1/3 because the scope ends up a 1MEG load in parallel with the 1MEG lower divider resistor giving 500k.
If you try measure signal on a gain stage's plate, it will get loaded down somewhat.

I realized the 100:1 probe is actually a bad idea. AC-coupled, a scope's input cap will charge to the full DC potential regardless of the probe attenuation, if it's 1x, 10x, 100x because the cap is ahead of the scope's 1MEG front-end resistor. This is a trap.

If you have any drama with the Leader, check it's PSU rail voltages are OK and then I go after the old Marcon caps.

If the cap can charge via the probe attenuation, it can equally discharge via that path.
As soon as you have done your test, clip the probe tip to the 'scope ground, & it will discharge harmlessly.

[jonpaul]

look for the P6006..6009 1960s X10 designed in vacuum tube era. Rated at 600V.

https://w140.com/tekwiki/wiki/P6006

as in film "2001: A Space Oddssey"
Can easily probe DC or AC at the plates of amplifiers.

Low cost as so big and old!


Jon

[vk6zgo]

look for the P6006..6009 1960s X10 designed in vacuum tube era. Rated at 600V.

https://w140.com/tekwiki/wiki/P6006

as in film "2001: A Space Oddssey"
Can easily probe DC or AC at the plates of amplifiers.

Low cost as so big and old!


Jon

I used them a lot in the old days, teamed with a 545B & 1A5 plugin.
We didn't "need no steenkin' AC coupling"----the 1A5 "offset" could handle anything we threw at it!