metaruss

Dad’s Williamson Amp – Description

BackHistoryDescriptionSchematicsPerformancePictures

Design


New power supply

The circuit is a classic “Williamson” design, based very closely on the circuit published by D.T.N. Williamson in Wireless World in 1947 and again later in 1949. The key differences are the use of 6BL7s in the driver stage (6SN7 is typical here) and 8417s setup as pentodes in the output stage in place of KT66s strapped as triodes. The original B+ supply provided 540V, which could deliver nearly 100W of power per channel. The new power supply delivers around 400V, reducing the maximum power output to between 40W and 60W (depending on what output tubes are used).

Input Stage

The input stage is 1/2 of a 6SN7GTB (section 2) twin triode, self biased at about -2V for about 4mA of plate current.  The plate voltage is about ?V and is directly coupled to the phase-splitter.

Phase Splitter

Like my Thomas PA, this amplifier uses the split load (or “concertina”) phase splitter using the other half of the above mentioned 6SN7GTB twin triode.  The grid voltage is set to around ?V, resulting in about ?mA of plate current.  The concertina, while well balanced when unloaded, has rather poor output impedance matching when faced with a difficult load to drive.  In this case, it does not have to drive to output tubes directly because it is capacitively coupled to a driver stage via a pair of 0.22uF film capacitors.

Driver Stage

The driver stage uses both halves of a 6BL7GTA twin triode, one for each phase.  They are self biased at about -?V for ?mA of plate current.  The plate voltage is about ?V and is capacitively coupled to the output stage via a pair of 470nF film capacitors.

Output Stage

Dad's Amp - 8417s - Close
The output stage used a pair of RCA 8417s with fixed bias.  When 540V of plate voltage was present in the original amp, the plate current would have been set to around 100-125mA for nearly 100W of power into the 3.3k output transformer.  The current power supply provides around 400V to the plates, so output is reduced somewhat.  The 8417 is capable of of 60W or so into a 2.8k load at this voltage, which yields around 40-45W into these 3.3k output transformers.  When 6550/KT88 tubes are used instead, the amp can reach around 60W of output at clipping.

The 8417 has a reputation for instability when configured for fixed bias.  They also have a problem with the screen overheating (particularly the GE version) and arcing over.  The screen geometry is why it has to be set to a much lower voltage than the plate (300V).  This is why the new power supply has an adjustable, regulated screen supply and a bias adjustment with enough range to allow the use of other tube types.

At this time, the 8417 tube is extinct.  I have seen pairs of RCA tubes (made by Sylvania) selling for as much as $200.  I have seen the GE version for about $50 each.  The price hasn’t changed much in the past few years.  Most people probably have either dumped the equipment or converted them for use with 6550s or KT88s.  The 8417 was one of the last audio power tube designs of the tube era.  Despite its shortcomings, it was a comparatively easy tube to drive.  They were only made for a short time before RCA and GE stopped making tubes altogether.  This would have been a great tube for a company like New Sensor to bring back into production, but they are difficult to manufacture reliably because of the tight screen tolerances.  Current manufacturers have enough trouble making common power tubes for guitar amps reliably.

Power Supply

New power supply

The original power supply had two parts. On the main chassis was a 700VCT power transformer that also had secondary windings for the filaments.  This was used to power the screens and the preamp plate supplies.  There was also a bias tap used for the output tube bias supply.  The filament winding was supplemented by another 6.3VAC transformer, which showed signs of distress.  The output tube plates where powered from an off-chassis supply after my dad had burned-out a half dozen or so plate supply transformers.

The new supply uses a toroidal transformer from Antek (AN-4T360) to power all of the plates and the heaters.  The B+ supply is a virtual split supply using typical CLC “pi” filters.  Subsequent RC stages pull from that to supply the input and driver stages.  The screen supply uses two “Maida” style regulators with a high-voltage MOSFET to handle the heavy lifting.  The minuscule gate current requirements allowed some adjustability to be designed-in so that other tubes types could be better supported.  The bias supply uses a small Hammond 263AX transformer that provides a positive and negative 100V rail to better support future endeavors.

 

Modifications


Eventually, the topology will be changed from classic Williamson to something a bit more modern, possibly sporting a MOSFET driver stage for AB2 operation.  That is why the new power supply more elaborate than is really necessary for this design.

Search

Use the form below to search the site:

Still not finding what you're looking for? Drop a comment on a post or contact me.

Sites that interest me

A few highly recommended friends...

Meta