Not just an replacement, but an improvement!

Technical Bulletin: TB-06

PX4 CONFUSION RESOLVED
Written by: Jac van de walle

Last Updated: 07Aug-2017


Finding the right track with PX4.
.

At Emission Labs, when building prototypes of the PX4, we made a decision: We are not going to build them with a 4Volt, 1Ampere heater, because that is not enough to build a strong emission, long life, and (to some extent) abuse proof tube. The heater specification by OSRAM, was defined when AC electricity distribution was not very common. So amplifiers were often running from rechargeable batteries with a relay chopper to transform this up to the anode voltage, while the heaters were directly heated off the batteries. Today, power saving is less of an issue. So we choose for 1.5 Ampere heater current, so to have 6 Watt heater power available.

However, with our prototypes we just couldn't get the specifications right. When the working point was right, Gm seemed wrong, and when we adpated Gm, the working point became wrong. We blamed it on the change we made with the heater current. So we gave up on the PX4, as we had no intention to build tubes with only 1 Ampere heater.

Later, we found out we were on the wrong track with our conclusions.

A historical, confusing situation.

Some 90 years ago, the PX4 tube data sheet was generated, still used today, and there is nothing wrong with it. Yet it is subject to great confusion, leading to Chinese replica tubes being build with wrong data. I am not aware somebody ever noticed, until this technical bulletin was first published in 2015. At EML we tried the same thing as the Chinese, and made the same error. To understand how this happened, we have to go back to the days when this tube was developed in the 1920's by the MO Valve company England, and marketed under the OSRAM and MARCONI brands, beginning in 1928.

At first there was a low power version, with a 0.6 Ampere heater. Later the high power version with a 1.0 Ampere heater was introduced. During those days, the best available commercial tube tester was the English AVO MK1. Today, this tester is almost forgotten, but we have to stand still here, and look at this tester in detail. In those days of the PX4 development, it was "the" reference. Later, the Hickok company, had something which they called "English Test" on their larger testers, and it was inspired by what the English were doing. So we do need to understand what a "world standard" the English AVO Mk1 tester was, and it was only obvious they presented the PX4 data sheet with the settings for the AVO Mk1 in it. This was really a good thing, and they did it that way. I have a good Mk1 in my collection, it works perfect, and accurary is amazing. However it does test the tubes are very unusual settings. Actually Hickok tests the tubes also at very unusual settings, it's just Mk1 tests the tubes at 0V grid voltage. But also at 0V grid, and 100V Anode, the tube has factory specifications, and whatever they are, these can be used to test the tube.



AVO Mk1 tube tester. More about it on the www.jacmusic.com website

It seems almost certain, this was the whole reason for this very unusual test setting in the official data sheet, but today we can not ask those old Osram Designers any more, they have all passed away.

So we need to work with what they left us. That is the historical data sheet, and the historical tubes which collectors own. The result was: Ra=830 Ohms, and Gm=6mA/V, but you must keep in mind, this was specified at Ua=100V and Ug=0V. So take good note of this simple combination of numbers, as written in this sentence, and from here the mystery tour begins....

How the confusion continued

Unfortunately, the datasheet does not give Ra and Gm at the normal operating points. It's simply left blank. Now you don't absolutely NEED to know, because one can indeed determine the desired operating point by using the tube curves only. However historical PX4 is so rare, I don't think any amplifier designer would use an historical PX4 tube, and the historical datasheet, to design his products, and then tries what the replica tubes are doing in his amplifier. Though it would of course be the best way. To get hold of a really fine, historical PX4 you need to know a collector who leaves you his treasure for experiments. Very unpractical of course. So instead of that, they just take a Chinese tube, bias it at 250V, 48mA and there you are: it works. But does it really...?

We have some lovely misunderstandings on the PX4 replica market today. So you will see the Chinese have made a PX4 with 6mA/V because they copied that number from the Marconi datasheet. But... for the working point for this, they obviously thought 100V Anode and 0V grid is not a good idea. So for the working point they used the datasheet recommendation. So there you see: 250V/48mA and Ra is recommended 830 Ohms, so that was what they copied. Only where is Gm? Just look elsewhere and you see: Gm=6mA/V at 100V anode and 0V grid. So now they had Ra, Gm, and a recommended working point.

The mistake

If you can see the mistake already, well done! But many are unaware Gm is a function of the bias point. And not just a little bit. So the Gm data at 100V anode, is totally another at 250V anode. You see here, what plain copying of other's data sheets can lead to. Even so, the Chinese were not the only one, making this mistake. The weird thing is, the Chinese "PX4" with it's funny parameters, was used by many amplifier designers, and nobody ever found out, the parameters were in fact wrong. For this, you have to look carefully at the OSRAM Data sheet, and you will see there, Gm is NOT defined as 6mA/V @ 250V/48mA as the Chinese did it. It is 6mA/V @ 100V/65mA as OSRAM did it. And that is a MAJOR difference, and it would be a big mistake of course to ignore this. Any PX4 like this to my opinion should be called a PX3 or a PX5 but not a PX4.

Let's begin at the beginning

We have to admit it, we made initially the same mistake at EML, but then of course when verifying the prototype point by point, versus the original tube curves of OSRAM, the prototypes appeared unsatisfactory. They they were was not really bad, there is always some tolerance of course, but we just could not make a good bogey tube. We even stopped the PX4 project initially.

I am sure you already found this text confusing, and it is indeed. So let's begin at the beginning, and make an inventory in table form of what we know, and what we do not know about the PX4. That will result in the following table, where the red boxes are simply never filled in by OSRAM. I repeat, the background for this, to my opinion for 99% sure, was the AVO MK1 tester.

So here is the inventory of known data. These is all that was published by Osram: (here the original data sheet)

OSRAM data table

Original Osram Specifications
Anode Voltage
300V (Max)
250V
200V
100V
Grid Voltage
-42V
-32V
-28V
0V
Anode Current
50mA
48mA
25mA
65mA
Anode Dissipation
15Watt
12Watt
5Watt
6.5Watt
Bias Resistor (AC heated)
900
740Ohms
1200 Ohms
Load Resistor (Ra)
4000 Ohms
2400 Ohms
4500 Ohms
Gain (Mu)
Left Blank
Left Blank
Left Blank
5
Transconductance (Gm)
Left Blank
Left Blank
Left Blank
6 mA/V
Plate Impedance (Rp)
Left Blank
Left Blank
Left Blank
830 Ohms
Estimated power Output
Left Blank
Left Blank
3.5 Watt

So here is a problem for those tube Manufacturers that want to re build the PX4. Some of the data is missing as you can see. Moreover in the original datasheet, it is indeed marked that Gm , Rp and Mu are defined at 100V anode and 0V grid, but this is written there very confusing. So some day, somebody may decide a PX4 has Rp = 830 Ohms and Gm = 6mA/V. Then this needs an operating point, and there is one for 15 Watt and one for 12.5 Watt. Then combine this 15 Watt operating point, and the result is like in the next table. Now of course you are free to do so, but such a tube differs from to the OSRAM PX4, and that to my opinion that is too far away from a PX4 to call it a PX4. I will explain this more detail, why I think so.

The Chinese result is published like this on their website.

 
 

Historical OSRAM Specifications in brown.

Reconstructed data in GREEN

Chinese PX4
Specifications Published as numbers:

Chinese PX4 Specifications, as I have extracted this from the Chinese tube curves, as they published those.

Note the difference in RED)

Anode Voltage
300V (Max)
300V (Max)
300V (Max)
Grid Voltage
-42V
-42V
-42V
Anode Current
50mA
50mA
50mA
Anode Dissipation
15Watt
15Watt
15Watt
Bias Resistor (AC heated)
900
900
900
Load Resistor (Ra)
4000 Ohms
4000 Ohms
4000 Ohms
Gain (Mu)
4.9
5
5
Transconductance (Gm)
4 mA/V
6 mA/V
2.33 mA/V
Plate Impedance (Rp)
1233 Ohms
830 Ohms
1700 Ohms
Estimated power Output
4.5 Watt
4.5 Watt
It can be seen from this table, there is great confusion indeed. We have so say it, this is not caused by the Chinese factory, but rather by OSRAM, not publishing numerical data from the beginning on 1928.

On a Chinese website, there is a graph they made with the Sofia tester, and they write this is made with AC heater. I do not think so, because I have the schematics of the Sofia, and I own two of them, and they have DC heating, and no option for AC heating at all. They don't have the option in hardware, and not even a correction in software. So, ANOTHER mistake. This doesn't matter a great lot for a tube with -50V grid voltage, but you are indeed measuring the tube at 2 Volts offset. (Half the heater voltage). Just want to bring this up here, and do the math precisely. Some numbers on that graph were hard to read, so I photo shopped the revised grid voltage in the graph, including 2 Volts compensation for the DC heating. And now we are getting something, and the result looks like this:


Chinese PX4 on Sofia tube tester, bias point marked by myself.


Chinese PX4. Reconstruction of Gm Data


Chinese PX4. Reconstruction of Rp Data

The gain (Mu) can be calculated, it equals: Mu = Rp multiplied by Ra.

Chinese PX4 Specifications reconstructed from the above tube curves:
Anode Voltage
300V (Max)
Grid Voltage
-51V
Anode Current
50mA
Anode Dissipation
15Watt
Bias Resistor (AC heated)
900
Load Resistor (Ra)
4000 Ohms
Gain (Mu)
3.9
Transconductance (Gm)
2.33 mA/V
Plate Impedance (Rp)
1700 Ohms
Estimated power Output
4.5 Watt

To prove the above graphical construction is the right method, I will do so with the OSRAM data sheet, and reconstruct the values for Gm, Rp and Mu, at 100V Anode and 0V Grid. The result should be: Rp = 830 Ohms and Gm = 6mA/V

 

As you can see, I do not fully find 6mA/V but only 5.4 but it has to be said this method of only going down vertically (and not going up as well) in the chart, it gives some small error. However constructing is not possible otherwise in this way, as there is no data in this chart for positive grid voltage. (Any curves for positive grid voltage may not be just be penciled in, as dynamical tube data changes suddenly some 10% above +0Volt. You can check that yourself with curves of some triodes that do present positive grid voltage curves indeed) So, finding 5.4 instead of 6 proves just the graphical construction method is correct, and we know the missing 10% is from not having the positive grid voltage data available. We don't follow up on this here, as there is no need.

What are going to do: It is possible to construct the data from the normal operating points though, so at 300V and 250V. This will be done in the next charts, and this is going to lead to very precise Gm data, as needed.

Finally completing the OSRAM data sheet after 90 years :) .

We can just re construct the missing data, by using the tube curves.

 

Original Osram Specifications in brown
Missing data reconstructed in green, from OSRAM curves
Anode Voltage
300V (Max)
250V
200V
100V
Grid Voltage
-42V
-32V
-28V
0V
Anode Current
50mA
48mA
25mA
65mA
Anode Dissipation
15Watt
12Watt
5Watt
6.5Watt
Bias Resistor (AC heated)
900Ohms
740Ohms
1200 Ohms
Load Resistor (Ra)
4000 Ohms
2400 Ohms
4500 Ohms
Gain (Mu)
4.9
5.6
4.6
5
Transconductance (Gm)
4mA/V
5mA/V
3.2mA/V
6 mA/V
Plate Impedance (Rp)
1233 Ohms
1133 Ohms
1448 Ohms
830 Ohms
Estimated power Output    
3.5 Watt

 


Subscribe or unsubscribe our Mailing List
© EMISSION LABS . All Rights Reserved.