Allright, first post on the forum!
Anyway, time to jump right in. This is some work in progress I'd like to share with you guys & ladies. I’d like to build a simple hybrid headphone amp and the SSMH is really ideal for my purpose. I'm having some fun adding minor tweaks to the SSMH's design to make it a bit more flexible for my build. It's basically one way to bring the "Starving Student" back to the SSMH. Since the 19J6 is a bit scarce, I'm currently using the 12AU7-modded SSMH-schematic from Dsavitsk and the_equalizer as a baseline. With the appropriate adjustments, the SSMH can actually run with E88CC/6DJ8's and 6J6's, next to the already existing 12AU7-version. In fact, the standard PSU can run any reasonable 6,3V or 12,6V preamp tube with a current between 0 and 350 mA.
As I’m not allowed to upload pictures within the first few posts, I’ll just have to explain per keyboard. If you want to follow, get a schematic of the 12AU7-style SSMH by the hand.
Basically I used the same schematic, with only a few minor tweaks. Say I want to use a couple of E88CC’s instead of the 12AU7’s. Both are a dual triode but have different pinout’s. The 12AU7 has 2 heaters (one for each halve, rated at 6,3V at 150mA) internally in series. The heater has an accessible center tap so the heater can be externally configured in series (12,6V/150mA) or parallel (6,3V/300mA). The E88CC actually has the exact same heaters (2 times 6,3V at 150mA) but are internally connected in parallel. This means that it only can run in 6,3V mode at 300mA.
The first problem is the lower heater voltage of the E88CC and the 6J6 being 6,3V. So I added a couple of LM317HV constant current sources (IC + one resistor) between the mosfet sources and the tube heaters (a bit like the Sijosae MHHA v2.0 or the Bravo Audio V2 but at 48V). This decouples the heater voltage from the mosfet source voltage. The heater can then be configured according to one's wishes (6,3V or 12,6V) independent of the mosfet bias voltage. This is akin to putting a beefy power resistor in series with the heater to get a higher voltage drop. You can for example put a 21 ohm resistor in series with the E88CC heater and it will be equivalent to a 12,6V heater at 0,3A. The difference with a CCS is that the power resistor will pass AC voltage variations while the CCS will actively block most of the AC voltage variations, the constraint being the constant current draw. Since the Bravo Audio V2 is a working amp, I assume this tweak will certainly work, provided you use the HV-rated version of the LM317.
Now here’s where things get experimental. The second problem is that the current in these lower voltage heaters is a lot higher than your typical 150mA of the 12AU7. 300mA and 450mA are typical values for the E88CC respectively 6J6. So one mosfet carries the full heater current. Some tube amps use a string of preamp heaters in series as a cathode resistor to bias the output tubes of a push pull stage. This means you can e.g. have a couple of E88CC heaters in series giving you a compound heater of 6,3V - 12,6V - 18,9V - 25,4V at the same current of 300mA. Each output tube then gets 150mA of this heater current. The same concept can be applied to the SSMH. You use one or more tube heaters in series and the current then splits up into the mosfets for the left and right output stage. This can be done by having the heater shared between the left and right channel. The PSU can now be dimensioned at half the current of the traditional SSMH.
As an example, for a cool running SSMH I could use the heaters of a single 12AU7 in series, which gives me 12,6V at 150mA. Then I split the current into 2 times 75mA with each LM317HV configured as a 75mA CCS-load for it’s respective mosfet source follower. This can still be enough for some 90mW or more (e.g. 53mArms into 32 ohm) into the more sensitive medium to low impedance headphones while simultaneously halving the heat production to 7.2W (48V * 0,15A) instead of 14.4W (48V * 0,3A). Of course, nothing stops you from running the heater in center tapped 6,3V mode at 300mA and push 150mA per mosfet as in the original SSMH, while still using only one tube.
The following should be feasable for the original Cisco 48V PSU:
- Single 12AU7 at 150mA with shared heaters
- Single or dual 12AU7 at 300mA with separated heaters - this one contains the 12AU7-mod by Dsavitsk & the_equalizer
- Single or dual E88CC/6DJ8 at 300mA with shared heaters
And for a more powerful 48V PSU:
- Single or dual E88CC/6DJ8 at 365mA with shared heaters
- Dual 6J6 at 450 mA with shared heaters - possibility to get the authentic SSMH 19J6 tone at a very low tube price
The next logical step could be to DC-couple the mosfet to the tube plate. So on the schematic, remove C2,R2,R4 and C4,R8 and R10 and attach the resistors R3 and R9 directly to the tube plate and done. You can then play with the plate load resistors and the cathode bias resistors of the tube to find a good quiescent point of the tube while simultaneously biasing the mosfet for the most optimal voltage.
These small tweaks should actually have little impact on the sound quality (except for fooling around with the plate load and cathode bias with a DC-coupled mosfet) while bulletproofing the amp for first time builders. Provided you solder the current setting resistors of the CCS correctly, then too high a mosfet bias cannot burn out tube heaters.
A tone altering tweak could be to add a plate CCS to the tube. This would compromise the tube swing as any active CCS uses a bit of voltage to operate. So make it passive. We already have a mosfet source follower hanging up there. Split the plate load resistor into 2 equal resistors and add a capacitor from the mosfet source to the midpoint of the two resistors. The mosfet source follower is a unity gain stage so couples all the AC-variations back to the midpoint of the two resistors. The lower resistor sees a constant voltage drop and a constant current. This is called a bootstrapped current source and requires only one extra resistor and capacitor.
The schematics are still experimental but look promising. The DC-coupled version actually looks simpler than the 12AU7-modded SSMH. I’ll upload ‘em as soon as they are trustworthy for some fooling around on my breadboard. In the meantime I’ll be happy to answer any questions. And if this whole contraption doesn’t work: I’m GlowGale, pleased to meet you. And I like subtile first posts.
Greetings from a real starving student,
Edited by GlowGale - 10/11/14 at 8:51am
- Had to remove the single 6J6-option due to the common cathode between the triodes