Replacing Valves With Semiconductors

The basic unreliability, limited life and/or unavailability of valves has led to several attempts over the years to directly replace them with solid state equivalents in existing equipment. The first instance of this that I will look at is the so called "Fetron".

The Fetron

In 1970, a company called Teledyne Semiconductor started on the development of the "Fetron", a solid state replacement for various widely used valves.

At the time, valves were still used by the millions in telephone networks and other professional equipment and it would be greatly advantageous in terms of reliability and power consumption if the valves could be replaced by electrically equivalent plug-in solid state devices. This was the spur for the development of the Fetron.

The Fetron itself normally consists of two junction FETs connected in series as shown in the diagram below.

In this example, the TS6AK5 Fetron is used to replace the 6AK5 valve, a type used extensively in telecommunications equipment. As of 1972, Fetrons were also being made to replace the 12AT7, 12AX7, 6AM6, 6CB6A, 6EW6, 6JC6, 407A, 408A and others.

Fetron construction

Physically, the Fetrons were constructed on a ceramic substrate using standard hybrid thick film techniques. Other components, such as resistors and capacitors, were often included on the same substrate.

The substrate is soldered to a header. The connections on the chips and substrate are gold bonded to the ends of the pins. A protective case is cold welded to the cap and filled with nitrogen to protect the device. Various stages in the construction of the Fetron are shown on the next page.

It must be noted however, that the Fetron designs are not an exact replacement for the valve. The manufacturer emphasised that Fetrons must be extensively tested in the intended equipment to indicate the level of compatibility. To facilitate this, the TS6AK5 for example, was made in three versions that covered most applications.

Performance

Below is shown the anode curves for a particular type of TS6AK5 compared to the valve.

As you can see, the curves are not the same. In fact, the Fetron curves correspond more closely to an "ideal" pentode. As such it gives higher gain and had a higher anode resistance. It also had lower noise than the valve, and lower distortion, as did the others in the range.

The difference in the curves between the Fetron and the valve are more striking in the case of a triode, as can be seen below.

Despite the difference shown, the Fetron has superior performance to the valve in most circuits.

Usage

The Fetron was used mostly in telephone carrier and repeater equipment but kits were also available for certain high-end service equipment such as the Hewlett Packard HP400 AC VTVM and the CM module for the Tektronix 500 series of oscilloscopes.

One can imagine the benefits of using the Fetrons in a VTVM would make it a far easier instrument to use, ie, no warm up or settling time, no drift, and no heater induced hum.

The Fetron's main benefit however, was in equipment that was in continuous use, where it was estimated that it would pay for itself in about six months. Fetrons were given accelerated life tests with the prediction that they should last about three million hours, or over 300 years continuous use. This compares to a life of 50,000 hours for a typical high reliability commercial grade valve, or one to ten thousand hours for a consumer grade one.

However, the high cost of the Fetrons, about $12 at the time, meant that their use was pretty much confined to the commercial and military field. They were used, however, in some "Mesa Boogie" guitar amplifiers to combat hum and microphony problems.

A search of the internet reveals that there are still some for sale, if anyone wants to experiment. Hugo Holden, in 2007, actually made a radio using Fetrons. Western Electric also produced some similar FET based valve substitutes called HIN, or "Hybrid Integrated Networks", but only for use in their own equipment.

The PL802 replacement

The PL802 was a valve designed for video output and has the incredibly high transconductance of 40 mS. This valve was used in European colour televisions, notably the Philips K7, K70 and K80 chassis models, in the 1960s and 1970's. It had become hard to obtain for replacement purposes by the early 1980's. Various other valves could be used in most instances for replacement but apparently they were not successful in these Philips chassis. This prompted several manufacturers to come up with a solid state replacement which was satisfactory.

Above is a photo of one of these solid state PL802's and next to it is a pin-out diagram showing the circuit. The 56Ω resistor is the large power resistor in the photo. This substituted for the heater, necessary because of the series connected heater strings universally used in European televisions at the time.

Substitutes for older valves

In the 1990's to the early 2000's Tom Burgess from Arkansas advertised a transistorised substitute for early battery valves in "Antique Radio Classified" magazine. These were in the form of a wafer that fitted over the pins of a low emission or burnt out valve, making contact with the pins.

Below is an illustration from one of Tom Burgess's adverts. As you can see, different sizes were made to suit different battery triodes. The circuit was very simple, as can be seen from the diagram at the right. The pin numbers shown correspond to a UV or UX socket. An article with the circuit was published in the June, 1995 issue of Antique Radio Classified. This gave the values for the resistors and capacitor, and the type of transistor used.

Unfortunately, at the time I put this issue away in a safe place for easy reference, thus losing it for all time. A search of the forums reveals that they didn't always work very well, and the advert advises that "Performance will vary between radio models".

In particular, they didn't work too well in neutralised RF amplifiers and there was no volume control via the filament rheostats. In the February 1996 issue of Antique Radio Classified is an article describing a JFET based tube substitute that could be controlled via the rheostat. See the circuit below.

The source resistor is a photocell upon which light from the lamp falls. The lamp is controlled via the normal rheostat in the radio. For this to be effective, the lamp should be close in voltage and current rating to the original valve filament. A 6.3 volt 250 mA bulb would be ideal to replace a UX201A for example. The photo-resistor is given as a Mouser Electronics 338-76C348 or equivalent but these may be difficult to source now. The assembly must be shielded from ambient light of course. The FET is given as a 2N3819 or 2N5457. Other types may well be satisfactory. The HT is reduced to 18 volts, conveniently supplied from two 9 volt batteries.

Ray Bintliff of Antique Radio Classified tried two of these in a Crosley model 51 with good results.

My own experience

Back in 1996 I was given a 1920's home made three valve radio with a horn speaker and the instructions "Please can you get this going".

There were no valves and, as the owner wasn't that fussed about originality, I considered making some transistor substitutes as per the Tom Burgess design. The radio used a Reinartz type detector, followed by two stages of transformer coupled AF. The usual RF choke that is between the detector anode and the first AF transformer was absent. See diagram below.

After some experimenting I found that V2 and V3 could be substituted with the Tom Burgess transistor circuit.

The following values were found to work well: For V2; C1=10nF, R1=1MΩ and R2=2.2kΩ. For V3; C1=47nF, R1=1MΩ and R2=47Ω.

For the transistors, I used BC547's, and 18 volts for the HT. However, I couldn't get this circuit to work in the detector socket (V1) at all. It didn't detect very well and it certainly wouldn’t regenerate. So I tried variations on the FET circuit, finally arriving at the circuit below:

In theory, just a JFET by itself should have worked and it did but the circuit with the added source resistor and germanium diode worked better. I did find that a 2.5 mH RF choke added in the circuit in the position shown was necessary for proper regeneration. The HT was 18 volts, the same as for the amplifiers. The FET I used was a 2N5459 and the diode can be just about any point contact germanium type, such as an OA91.

I handed back the radio with the above substitutions and with two 9 volt batteries attached to the base board for the HT. The LT+ terminal was connected to earth and the filament rheostat had no effect on the circuit operation. I was able to demonstrate the radio operating but had to explain the limitations of a primitive 1920's receiver. I'm not sure if the owner ever bothered putting up an outdoor aerial for it, or arranging an earth.

I've always intended to experiment further with solid state valve substitutions and have considered using high voltage power MOSFETS which are readily available. In particular, I would have liked to make substitutes for expensive or hard to get valves, like the 45. However, the circuit would need some sort of negative feedback to make it behave like a triode, and the high input capacity of power MOSFETS would have to be circumvented somehow. I don't have time to do this work now.