Heathkit VT-1 Vibrator Tester.

With my other vibrator equipment, it was only natural that a commercially made vibrator tester should be added to the collection. The Heathkit VT-1 appeared in 1953 in the U.S. and was a popular piece of test equipment.
I already have built a vibrator tester as described here. It operates on very different principles and requires a CRO to observe the vibrator condition. My design allows one to check:

Vibrator testers designed for the average radio serviceman are simpler because such detail is not required. The serviceman just wants to know if a vibrator is good or bad, in the same way a valve tester shows good or bad valves. If a vibrator appears faulty, he simply gets one from stock and replaces it. Remember, unlike the modern enthusiast who has the time and motivation to examine details and repair vibrators, in the valve era, repairs were done for profit, and therefore minimum time for each repair was necessary.
Hence, the Heathkit VT-1 resembles a valve tester, with an array of sockets for different types of vibrator that may be encountered, and a meter displaying the condition with minimal interpretation.
The VT-1 is essentially a vibrator power supply with a transformer, load resistor, and output volt meter, and so operates the vibrators as they would in a piece of equipment. When non-synchronous vibrators are tested, the user has to provide a rectifier valve which is plugged into socket "A". This could conceivably be the rectifier from the car radio if it is a 6X5 or 0Z4.



How to Operate the VT-1

The VT-1 requires an external DC power supply; variable from about 4V to 6V, and with a current capacity of about 4A. Heathkit recommend using a battery eliminator of the type used for testing car radios, and not surprisingly, such a unit (BE-4) is part of the Heathkit inventory.
The vibrator under test is plugged into the appropriate socket, and the "shunt drive - separate drive" switch is set appropriately depending on vibrator type. For example, Oak types have a separate contact for the driving coil, whereas most Mallory types are shunt drive, with the driving coil connected across one set of switching contacts.


The BE-4 was recommended for the power supply to operate the tester.

If the vibrator is non synchronous (otherwise known as an "interrupter" type), a rectifier valve must be plugged into the octal socket. The valve socket is wired to take a 6X5 or 0Z4. If the vibrator is synchronous ("self rectifying") the valve is not inserted.

Now the vibrator can be tested. Push the "Set input volts" button, and with the battery eliminator at minimum volts, gradually increase the supply voltage to the "Start" line on the meter. The vibrator should consistently start at this lower than normal voltage.
If it passes the start function, press the "Set input volts" button again and increase the supply to the "Test" line on the meter. The vibrator is now operating at normal supply voltage, and the condition can be read off the meter scale.
Two scales are provided because the output voltage differs between synchronous and non synchronous types with a valve rectifier. This is because there is a voltage drop across a valve rectifier, whereas the contacts of a synchronous vibrator have no voltage drop.


 



Limitations of the VT-1.
Firstly, the VT-1 can only test 6V vibrators. 12V is not catered for. However, it must be remembered this instrument was made for the U.S. market where automotive electrical systems were standardised at 6V until 1953. Nevertheless, this limitation would become problematic during subsequent years, as more 12V car radios came in for repair. Mallory had upgraded their 6VT1 tester to accommodate 12V in 1953 to the 12VT1D model (shown further down). It's a pity Heathkit didn't do likewise.
As an Australian user, I can't complain too much about this, but here 12V electrical systems were common due to the popularity of English cars. The VT-1 was no longer listed in the Heathkit catalog by 1957.
 


By mid 1956, Heathkit acknowledged the 6V limitation of the VT-1.

Secondly, the range of vibrator sockets does not include the UX-6 base used with many synchronous types, and the UX-7 base used with split reed types.
In fact, despite the array of sockets, there is only one Australian made vibrator that can be tested on this instrument as wired; the Oak V5105 with UX-4 base. Provided sockets C and F are rewired, the Ferrocart PM237 and PM104 can also be tested.. Again, this is not meant to be a criticism, as the VT-1 was designed for the U.S. The 3 pin Delco socket is not included as it was only used for 12V types.


Two Australian vibrators that can be tested; at left is the Oak V5105, and at right is the Ferrocart PM237.

A limitation of testing non-synchronous vibrators is that the condition of the rectifier valve will also affect the meter reading. Therefore, if the output reading is low, the rectifier valve should be substituted with a known good one, before concluding the vibrator is weak. In this regard, the VT-1 can be used to also test the rectifier valve.

Another limitation is that the "one size fits all" meter scale does not accurately reflect the differences in duty cycles of different vibrator types.



The Design.
Interestingly, it appears that Heathkit has cloned the VT-1 from Mallory's 6VT1. Apart from minor differences in the two calibration pots, the circuit is the same.


Mallory's 6VT1 tester circuit. At right is shown the Mallory 12VT1D tester. Note how the Heathkit meter resembles that which Mallory have used. The 12VT1D includes the 3 pin Delco socket.


The Heathkit VT-1. Note the similarity.

How the Tester works.
There are essentially three parts to the circuit; 1) Metering, 2) Synchronous vibrator power supply, and, 3) Non-synchronous vibrator power supply.
A number of different sockets are provided which are:

The sockets B to F are simply connected in parallel with each other into the tester circuit. It is not possible to change the pin configuration, but in most instances this is standardised.


Mallory type 245A being tested in socket "B".

The "Shunt drive" / "Separate drive" switch is connected to sockets D and E. In the case of socket D, the centre pin is the reed connection for shunt drive types. For the Oak vibrators using this base, the centre pin is the driving coil, with the reed connection being the vibrator can. For this, and the 7 pin reversible variant of the shunt drive vibrator, which also uses the can as the reed connection, a grounding cup is provided for the socket. It connects to the chassis of the tester and the negative supply input. For more details on this reversible base, see here.


Testing a reversible synchronous vibrator (Mallory 725C). Note that no rectifier valve is used.

For socket E, the switching is actually unnecessary, as the pin 4 is vacant on shunt drive types. Whether 6V is connected to this pin all the time or not makes no difference. In fact, many car radios have pin 4 connected even though they are designed with shunt drive types. It allows one to use a separate drive type as a replacement with no modification.

Socket C appears to be obscure. Even with a reasonable collection of U.S. made vibrators, I do not have any fitted with all eight pins.  Looking through the vibrator catalogs does not show any. However, it does take the Delco 5 pin base used by some synchronous types. Interestingly, looking at the list of Mallory types that plug into this socket, as seen in the previous photo of the 12VT1 tester, we see 273D, 514 and 716.
These all use the 5 pin base. At this point it is doubtful that any vibrators were actually made with all eight pins.


Mallory type 245C being tested in socket "C".

Socket F is for the Delco 4 pin base. It is a UX-6 socket but with two of the pins (2 & 5) blanked off. A vibrator with the 4 pin Delco base will thus fit a UX-6 socket. In Australia, this base suits the common Ferrocart PM237 and PM238 used in Astor car radios. In fact, most of these radios did use a UX-6 socket.
However, because the AWA/MSP Oak equivalents of the PM237 and PM238 have six pins (V4010 and V4016), they cannot be tested in this socket.

1) Metering.
The meter displays the input voltage when the push button switch is pressed. Current flows from the positive supply input via the meter, the normally open switch contacts, and the 200 ohm calibration pot, back to the negative supply. The meter is not actually calibrated in volts, but simply as "Start" and "Test". "Test" is the normal running voltage and is therefore 6.3V. When calibrated thus, "Start" shows the input being 5V. The point of this is to determine how reliable the starting of the vibrator is. With shunt drive types, any wear in the contacts shows up with poor starting. It can be assumed that a vibrator that won't start at 5V has become worn or out of adjustment. I have seen some good shunt drive vibrators which will actually start with as low a 2V input.

For displaying the vibrator condition, the meter measures the 250V supply developed across the 8uF filter condenser. In this operation, the meter is connected to a voltage divider across the 250V supply. At the top end of the voltage divider is the 5000 ohm load resistor, and at the bottom end is a 400 ohm pot. The meter is connected across this pot which functions as a current shunt. Again, the meter is not actually calibrated in volts, but the result is displayed as a variation of good or bad.

2) Synchronous vibrator power supply.
A standard 6 to 250V vibrator transformer is used. The circuit is conventional except for one thing. In a typical circuit, the rectifying contacts switch each end of the secondary alternately to earth, so that the centre tap becomes the positive output. This is because the reed is earthed. In the VT-1, operation is reversed so that the centre tap provides a negative output. This is simply done by reversing the secondary connections. The reason for this arrangement is to accommodate a valve rectifier for non-synchronous testing, without having to provide any switching.
The 250V is filtered with an 8uF electrolytic condenser. Loading is provided by a 5K 20W resistor, to draw 50mA. Thus, the vibrator is operated under the same kind of load it would be in a typical car radio. This will show up any defects in the rectifying contacts, as well as poor duty cycle with the primary contacts in non-synchronous types. The buffer condenser is a Mallory "Plascap" rated at .006uF 1600V.

A potential problem is with reversible vibrators - if the vibrator is installed the wrong way round, the 8uF electrolytic will be exposed to reverse polarity. Therefore, the initial test must be done quickly, and if the meter shows a negative reading, the vibrator needs to be reversed.
Possible damage could be avoided by using a non polarised condenser, but in reality, a second or so of reverse polarity is usually harmless.

3) Non-synchronous vibrator power supply.
This is the same as above, except that now a rectifier valve is brought into circuit. This is plugged into octal socket A. The rectifier is a 6X5 or 0Z4. These have the same pin connections, except for the 0Z4 the heater pins are not used, since it is a gaseous type.
The rectifier plates are connected to the secondary of the transformer in the usual way. However, because the rectifier cathode is earthed, the transformer centre tap again provides a negative output. With a given transformer, rectified output voltage is less with a valve rectifier, than with a self rectifying vibrator, and the meter scale accommodates this.


My Heathkit VT-1.
Like most U.S. made equipment in my collection, this came via eBay. There were only two other bids and as usual, the postage cost was a lot higher than what the actual tester cost. Condition is very good, compared to some testers that have appeared.


Note the burned wire along the bottom.

It was noticed that at some time a shorted vibrator must have been plugged in, as the 6V feed wire had the insulation burned off for most of its length. All components were original. For the initial testing, I simply used the nearest 4 pin vibrator which was an ATR 1400. For the rectifier I found a new looking 6X5. A circuit breaker would have been a worthwhile inclusion to the design.
I noticed first that the "Start" and "Test" readings were reading low. With the vibrator in and happily buzzing away, the output meter read about half scale. The vibrator waveform was also examined on a CRO, and was as it should be. As expected, within a few minutes the buffer condenser started to fail. This was evident by an increase in input current, and a warming of the condenser.

The Buffer Condenser.
The value of .006uF is not a preferred value. So what to replace it with? It so happened I had some 1500V Philips types rated for pulse operation, .0036uF and .0024uF. In parallel, these create the .006uF I was looking for. Alas, in circuit something wasn't right. The vibrator wouldn't start easily, and when it did the waveform was awful. This was at odds with the original buffer, and temporarily reconnecting it provided normal operation again.
One characteristic of shunt drive vibrators, of which the ATR 1400 is an example of, is that starting is erratic if the circuit loading is not as it should be, and in this case the incorrect loading was due to insufficient buffer capacitance.

I had been curious that this tester is claimed to work with vibrators of 100 to 250 c/s. Also, not all vibrators have the same duty cycle. So, how could they specify a certain buffer capacitance? Strictly speaking, it's impossible with such a variation in vibrator types.
I can only assume that the Mallory 6VT1 (which this tester is a clone of), was intended only to test Mallory vibrators which are standardised at 115 c/s. The variation in duty cycle with Mallory types is not great enough to worry about for the purpose of testing.
Interestingly, when I measured the original buffer, it was close to .01uF. When I tried a new .01uF, it too worked correctly with the ATR vibrator. The ATR tested had a lower duty cycle than a new Mallory, so these findings are perhaps not surprising.


Testing a Mallory "Gold Label" 1601 in the 4 pin socket "E".

Anyway, the next thing to decide is what to do about it. The ideal value of buffer capacitance has been discussed elsewhere on this site. Where a compromise has to be made with its value, it is better to have more than less capacitance. Too low capacitance exposes the vibrator contacts to arcing, and can damage the transformer insulation. Too high capacitance exposes the vibrator contacts to excess current, and while arcing may be eliminated, the incorrect value does not stop contact erosion by electrolytic contact material transfer.

For the VT-1, I settled on the lowest frequency vibrator with the lowest duty cycle that will be tested which is the 100 cycle Oak. The real world buffer capacitance value that suited this vibrator characteristic was again .01uF. Mallory (115c/s) and Ferrocart (150c/s) types were also tested, and as expected, the buffer capacitance was slightly high with the Mallory and higher again with the Ferrocart. However, it was not so high to be problematic, unless the vibrators were run for days on end like that. It is true that I could have included a resistor in series with the buffer to reduce the effects of excess capacitance as described here, but for the purposes of testing, it's hardly worth it.


Burned wire replaced along with buffer condenser.

Socket C.
While this socket accepts the unusual Delco 5 pin vibrator base, the wiring did not match up with the types I have in my collection. This is because the socket was wired to suit the Buick type. Seeing as I have a quantity of vibrators using the Delco base, I rewired the socket to suit types such as Ferrocart PM104, Mallory 245C, and Delco 5039661. These have different connections to the 273D, 514 and 716 mentioned previously.


Close-up of socket C. The eight pins are not arranged in a circular formation. Base of a Mallory 245C is shown at right.

Socket F.
While attempting to test some Ferrocart vibrators, they would only work in half wave. Even ones which I knew had good contacts still gave a half wave waveform on the CRO. With an opened Ferrocart, it became clear what the problem was - the tester socket was not wired to suit. Transferring the wire from pin 6 to pin 4 fixed that so both contacts now worked. In retrospect, I could have simply bridged pins 4 and 6 to allow both socket connections. In reality, this socket is only likely to be used with Ferrocarts.

The 8uF electrolytic.
Despite its age and the appearance of something exuding from the seal, it did actually reform with a leakage current of only 1mA. A peak to peak ripple voltage check showed the capacitance and internal resistance was quite OK. I did try a new one just for comparison but there was no difference, so the original stayed.

Calibration.
Not having the instructions for the VT-1, I had nothing definite to go on, and both calibration pots were obviously in need of adjustment. However, an article in "Service" for November 1952 did describe the Mallory 6VT1. Unfortunately, it does not say anything about calibration.
As it's clear that the "Test" line on the meter is for normal running voltage, it should therefore be set for 6.3V, and this would be a logical adjustment for the 200 ohm pot. This is what I did, and it turns out the "Start" line on the meter corresponds with 5V, which is a likely figure.
The output calibration is somewhat arbitrary. Half scale for a new Mallory synchronous vibrator was a bit low, so I readjusted the 400 ohm pot for about 3/4 reading. As it happens, there wasn't much more adjustment available. Even with the pot at maximum resistance it still isn't possible to reach full scale.
 
 
Scale Output Volts
1 158
2 175
3 191
4 205
5 223
6 241
7 260
8 284
9 not tested
10 not tested


Further Improvements.
I regard the vibrator waveform to be an essential part of testing and it is tempting to install a set of terminals in the side of the tester to connect a CRO.  Also, as the range of sockets is limited, it is tempting to replace socket C with a UX-6 type. I'm loathe to modify commercially made vintage equipment, but there's any easy way that solves both problems.


Oak V5211 (65UH) split-reed vibrator being tested. Red wire is for coil supply. Terminals allow monitoring of waveforms.

I constructed an adaptor box which makes the VT-1 much more versatile. A diecast box was fitted with three sockets, one UX-7 for split-reed vibrators, and two UX-6 sockets; one for synchronous types, and the other for dual-interrupter types.
This plugs into socket B with a UX-5 plug. A separate coil supply wire is provided for separate drive vibrators. This connects to the positive terminal of the VT-1.
Also, terminals were provided for the primary and secondary transformer connections so that the waveforms can be observed. This feature can be be used with any of the sockets on the VT-1 in use (except socket B of course), as the primary and secondary connections of all sockets are in parallel. The terminals are also useful for testing wire connected types such as the Mallory 222.


Mallory 222 is a wire connected synchronous type.

The UX-7 base split-reed vibrator is tested the same as synchronous types. Both reeds are connected together. One of the UX-6 sockets is wired for dual-interrupter vibrators. A two way switch selects either set of contacts. Simply wiring the contacts in parallel would not be an accurate test since one good set of contacts will mask the others if they are worn or dirty.
Oak type V4006, as used in Ferris car radios, can be tested in either of the UX-6 sockets as its contacts are the same as the primary contacts of a synchronous type, or one set of dual-interrutper contacts. When tested in the dual-interrupter socket, the switch must of course be set accordingly.
Of course, the 6X5 must be plugged into socket A when testing dual-interrupter or V4006 types.
In the case of separate drive vibrators, because of the separate coil supply wire it is now possible to test six or seven pin vibrators of 4V, 12V, 24V, 32V, or any other voltage in this tester. The VT-1 is powered from 6V in the usual way, but the coil is connected to another power supply of suitable voltage. As the drive coil in separate drive vibrators is not connected to the transformer switching contacts, the contacts do not have to work at the same voltage as the coil.

Oak types V4010 and V4016 are shunt driven but with a separate drive contact. They are fitted with a UX-6 base. I have not provided for this type yet, but may do so in the future by adding suitable switching for one of the UX-6 sockets.

Incidentally, the quality of the UX-6 sockets I used is very poor. They were purchased from eBay from a Chinese seller. The contacts have no springiness in them and do a poor job of gripping the pins. In fact, the contacts would drop out of the socket until I splayed them apart more. Unfortunately, buying decent sockets from the U.S., such as those made by Amphenol or Cinch, means an exhorbitant postage cost. The UX-7 socket looks like it came from the same factory but seems to be a bit better.

A note on Split Reed types.
It is important to know that there are two types of split-reed vibrators. One is the usual self rectifying (synchronous) type such as Oak V5211. As with the standard synchronous types, the secondary contacts close later and open earlier than the primary contacts.
However, in the U.S. another kind became popular in the late 1950's. With this type, the vibrator is actually a dual-interrupter, or to use the U.S. term, a "Duplex" type. Common types are the Mallory 1701 or CDE 5722.
Both sets of contacts are timed the same in standard dual-interrupter fashion. Rectification is done with solid state or valve rectifiers. To understand why this type came into being, remember that there was a sudden change in U.S. automotive electrical systems starting in 1953. All the car manufacturers had switched from 6V to 12V within a few years. With many commercial vehicles fitted with new 6V communications equipment, a problem existed in that when the vehicle was traded in the following year, chances are there would be a 12V system to contend with. To retain the radio equipment would entail modification; installation of a new transformer and vibrator being part of it.
An ingenious way around this was to have a transformer with two 6V primary windings; each operating with their own vibrator contacts. For 6V operation, both circuits would operate in parallel. But, for 12V, the circuits would be in series. The vibrator drive coil would operate off the supply at the mid point of this series circuit. Thus, no change of vibrator or transformer would be required.

Therefore, these types should be tested as dual-interrupter types and not synchronous, despite the pin connections being the same. It is true they will self rectify, and will actually test OK in the VT-1 as a synchronous type, but this won't give a true indication of the second contact set since the current is much less than when the contacts switch primary current.
Furthermore, some types are marked with "6 or 12" volt operation. The coils are actually 6V, and the dual voltage marking simply indicates the vibrator does not need to be replaced when the circuit is switched for 12V operation. It does not mean the coil is suitable for both voltages! If the vibrator is used in a more conventional circuit, it is necessary to connect a resistor in series with the drive coil pin when 12V operation is required.

For reasons of efficiency and contact life, dual-interrupter and synchronous types should not be interchanged.



Voltage Dependent Resistor.
An observation I have made over the years is that a few types of vibrator transformer will generate excessive output voltage when the operating conditions are not "just so". It turns out the transformer used in the VT-1 is one of these. This became evident when testing a vibrator which was sufficiently out of adjustment so that the buffer capacitance was indadequate. This particular vibrator was a shunt drive type, and as is typical with this type, when presented with inadequate buffer capacitance, starting becomes erratic as does the reed vibration. The result is a rather awful looking half wave type of waveform. Combined with certain types of transformer, the output voltage contains very high voltage spikes. It is enough to arc over at rectifier socket pins! This was observed with the VT-1.


This waveform was so destructive I didn't want to spend too long photgraphing it. Arcing was evident at several points.

It must be remembered that as a tester, the VT-1 will have unknown vibrators plugged into it and these kind of operating conditions can occur. The problem is that if arcing is sustained, carbon tracking will develop in the socket wafers. This of course is conductive, and the sockets are ruined. Worse still, there's a good chance of a breakdown in the transformer. Obviously, some kind of clamp is needed to limit the output voltage under fault conditions. The danger occurs when testing non-synchronous vibrators before the rectifier has warmed up. Given the 250V per side rating of the transformer, a mains voltage VDR is an ideal choice. A type S20K300 was chosen. It clamps at 300V.


VDR is connected across one half of the transformer secondary.

This worked very well. Under normal conditions, no current flows through the VDR, and the waveform is as it should be. Under fault conditions, the voltage is clamped and no arcing takes place. About 6mA was observed to flow through the VDR. This was not an accurate measurement, given the waveform, but proves the VDR is conducting. The particular vibrator used for these tests was an ATR 1400 which was out of adjustment, so that that duty cycle was shorter than normal, and starting was erratic (it's a shunt drive type). Starting did not usually occur unless the 6V was rapidly applied several times.


With the VDR, the voltage is clamped to a safe level. It can be seen the vibrator is operating in half wave with a short duty cycle. With certain transformers, excess voltage is produced.


What the waveform was with an almost correctly working vibrator, when the vibrator did start properly. The duty cycle is shorter than it should be and the buffer capacitance is only just adequate for these conditions. The VDR appears open circuit under these conditions.



Home