[Amps] Alpha Seventy HV Meter Readings

[Jim Garland]

I've not looked at the Alpha 70 circuit, but from the discussion it 
appears that the HV is measured by putting a 1 mA meter in series with a 
high resistance voltage divider, with a shunt across the meter to give 
the correct calibration. To me, that suggests Bill's surmise may be 
correct: Alpha designed it badly in the first place, and for the reason 
Paul mentioned. One can too easily dump 4kV onto the meter switch and 
surrounding hookup wire when rotating the meter switch.

The method I prefer is to have a fixed voltage divider, with a high 
resistance on the upper arm of the divider and a low resistance on the 
bottom arm, the values of the resistors chosen to give a 1:1000 voltage 
division ratio. For example, in my last amplifier, I wanted a 3kV full 
scale meter reading. The voltage divider resistors serve double duty as 
the bleeder resistor for the power supply, so I use a 200K/100W resistor 
for the upper arm of the divider, and a 200 Ohm for the lower arm. With 
these values, 3000V on the top of the divider gives 3V at the bottom 
resistor.  The bleeder current is then 3000V/200200 Ohms or about 15 mA. 
To use a 1 mA meter, one then bleeds off 1 mA of this current by tapping 
into the 3V point on the divider, with an appropriately chosen resistor 
in series with the meter. That way, there's never more than 3V on the 
meter circuit and its switch. Calibration of the meter is done by 
adjusting the series resistor, which is large enough in value not to be 
finicky.
In my amp, I use a 100 uA meter, which has an internal resistance of 
about 2K ohms. The total series resistance (which includes the meter 
resistance) is about 30K, so I use a 50 K trimpot to get the calibration 
exact. In my case, the loading of the 30K on the 3V is negligible. With 
a 1 mA meter the 3V is loaded down a few percent, but that is easily 
compensated for by adjusting the series trimpot. The one disadvantage of 
this design is that one shouldn't ground the negative terminal of the 
meter directly to the chassis, but rather connect it to the B- line. 
Otherwise, there will be some interaction of the HV meter with the grid 
current.
73,
Jim W8ZR

On 1/20/2017 2:57 AM, Bill Turner wrote:
> That still doesn't explain the original error.
>
> I see two possibilities:
>
> 1. Alpha designed it wrong in the first place.
> or
> 2/ Some component changed value.
>
> So, which was it? Or was it something else?
>
> 73, Bill W6WRT
>
>
> ------------ ORIGINAL MESSAGE ------------(may be snipped)
>
> On Thu, 19 Jan 2017 20:55:32 -0500, W9AC wrote:
>
>> I must be the only person who didn’t know the Fluke DMM can source exactly 1 mA.  If I read the manual in the past, it probably didn’t register.  Not sure if the internal source is regulated as the battery ages.  Time to review the manual.  Anyway, that method works great.
>>
>> My favorite "grab 'n go" DMM is a Fluke 8060A from the mid-1980s.  I have three of them; two are still new on the shelf in the original boxes. It's not auto-ranging, so it doesn’t slow me down when taking multiple measurements.  I prefer to range it myself.  The 8060A's lowest DC resistance range is 200 ohms.  In that position, the Modutec meter reads 34 ohms versus my manual method that shows 32 ohms.  Because of various mechanical limitations, the Fluke is probably more believable.  In the 200-ohm range and with the leads applied to the meter terminals, it reads...almost full scale as Jim pointed out!   BTW, it's only the in the lowest resistance range that the 8060A sources 1 mA.
>>
>> To Bill's question, the 1% metal film resistors all measure on the high end of tolerance.  1% of 1-meg is 10K ea. for a total of 50K, but that still doesn’t come close to the 500K required change.  With the Fluke DMM applied, it does not reach full scale and is off by about half the discrepancy amount.  If the Fluke is sourcing very close to 1 mA, then the meter may be contributing to the error.  I would like to believe it except that an identical Modutec meter reads the same error.  But I have to believe it because the math doesn’t make sense otherwise.  The meter's internal resistance of 32 ohms in parallel with the 15K pull-down resistor is a miniscule resistance in the string.
>>
>> Next, I need to source exactly 1 mA and observe the result.  I may have two identical meters that read low.  Since the meter is used for HV and Ig, then grid current is probably reading a bit low.  Recall that my Alpha 70V uses the same components as the 70A, yet metering in the 70V's HV position agrees with the Fluke's HV probe.
>>
>> By the way, with the meter at 4/5 scale for 4KV of HV, that results in 0.64 watt of 1-meg resistor heat dissipation.  It looks like Alpha used 1-watt/1% metal film resistors.  My change brought the cold-end resistor in the string down to 500K from 1-meg.  The computed dissipation from that resistor is 0.32 watt.  I used a 1/2 watt/1% metal film resistor, slightly elevated from the PCB.
>>
>> Paul, W9AC
>>
>> -----Original Message-----
>> From: MU 4CX250B [mailto:4cx250b at miamioh.edu]
>> Sent: Thursday, January 19, 2017 4:46 PM
>> To: Paul Christensen <w9ac at arrl.net>
>> Cc: amps at contesting.com
>> Subject: Re: [Amps] Alpha Seventy HV Meter Readings
>>
>> Paul, your method of measuring internal resistance of the meter is just fine, though it's easier just to use a DMM and measure the resistance directly. I imagine you have a fluke DMM and these normally provide exactly 1.000mA in the resistance mode. Actually, you're killing two birds with one stone because you can see if your panel meter reads full scale.
>> 73,
>> Jim w8zr
>>
>> Sent from my iPhone
>>
>>> On Jan 19, 2017, at 12:12 PM, Paul Christensen <w9ac at arrl.net> wrote:
>>>
>>> Here's a quick update.  The math worked and the replacement resistor
>>> value of 550K results in a HV reading of 4KV.  This matches a Fluke DMM with HV
>>> probe.    The new HV meter divider string is now 4.55 Meg, slightly reduced
>>> from the 5 meg design.
>>>
>>> As previously indicated, the multimeter is a Modutec 1.0 mA DC
>>> movement.  I decided to make the effort and measure its internal DC
>>> resistance.  For the measurement, I first selected a series R and
>>> applied a few DC volts from a bench power supply.  The supply voltage
>>> was increased until the meter read full scale (i.e., 1.0 mA DC).
>>> Next, I shunted the meter terminals with a 2K pot as a rheostat and adjusted it until the meter read exactly 1/2 scale.
>>> At that point, current is evenly divided between the meter coil and
>>> rheostat.  Finally, I removed the rheostat from the circuit and
>>> measured its resulting resistance.  The answer is 32 ohms.  That 32
>>> ohms consists of the meter coil and any other internal resistance inside the meter enclosure.
>>> So, terminal-to-terminal, DC resistance is 32 ohms.
>>>
>>> Back to the Alpha 70A: In addition to the original five, 1-meg HV
>>> metering divider resistors, a 15K resistor shunts the meter terminals
>>> when the multimeter is in the HV position.  At least with this
>>> amplifier, the 15K resistor is definitely NOT being used to sample current for the HV reading.
>>> The resistor is well more than 10x the meter's internal resistance.  I
>>> again verified HV multimeter accuracy by comparing results with the
>>> 15K resistor in and out of the circuit.  As expected, there's little
>>> change in deflection since the meter's internal resistance is swamping the 15K shunt resistor.
>>>
>>> In looking at other amplifier schematics from Ameritron, Heath and
>>> others, most use the same HV metering configuration: the multimeter
>>> coil is shunted with a resistor when the multimeter switch is in the
>>> HV position.  Now, it's possible that in those amplifiers that the
>>> resistor may be used as a sample which has an additional benefit of
>>> stabilizing readings -- but only IF the meter coil has a high internal
>>> DC resistance that approaches the value of the shunt.
>>>
>>> In the Alpha 70 series, the 15K meter shunt in the HV position is
>>> performing only one function:  The resistor is floating-down 4KV of
>>> high voltage that would otherwise be present at the moment the
>>> multimeter switch is engaged in the HV position.  Otherwise, with no
>>> meter current, the full 4KV supply potential appears on the multimeter
>>> switch, which may lead to arcing in addition to it being a safety
>>> concern.  Well, it's already a concern in an openly exposed amp.  But
>>> few folks would intuitively think that the full HV potential could
>>> appear on a small multi-meter switch.  With phenolic used as the
>>> insulating material on most multimeter switches, then all the more
>>> reason to have the resistor in place.  With the shunt resistor in
>>> place on the HV supply side, that level is brought safely down on the cold end of the HV divider string.
>>>
>>> This is probably way more than anyone wants to read, but I wanted to
>>> close the loop with my findings and report a solution that now results
>>> in accurate HV readings.
>>>
>>> Paul, W9AC
>>>
>>>
>> _______________________________________________
>> Amps mailing list
>> Amps at contesting.com
>> http://lists.contesting.com/mailman/listinfo/amps
> _______________________________________________
> Amps mailing list
> Amps at contesting.com
> http://lists.contesting.com/mailman/listinfo/amps