Bluebell Audio "WE91" Single Ended Valve Amp Monoblocks
The Loftin White 2A3 has pride of place in my system. After a couple of decades of solid state grain I was enjoying the wonder of direct coupled RCA 2A3s (dualplates admittedly). I'm having to use my old Exposure pre amp to control it unfortunately but it sound's great and I'm really happy with it. Why do I need anything more?
Well, because I'd become hooked... without realising it. That I was happy with the Loftin White was irrelevant, I needed to smell solder fumes. Nobody warned me how addictive this hobby could become. And I was now aware of how many valves I could try in so many configurations; single ended of course!
During the course of one of my conversations with Philip at Bluebell Audio, Philip had raved about a pair of 300B monoblocks based on the old WE91 circuit which he had just finished after sitting incomplete on his shelves for a few years. Over the following few phone calls Philip's enthusiasm for the WE91s didn't wane at all. Here, it seemed, was a 300B SE amp that could do bass and treble as well as the renowned midrange... Here was my next project.
The actual incarnation of the WE91 is Philip's so it's not appropriate to post it here, but it's very similar to the Angela circuit at www.Angela.com. If you speak to Philip he'll no doubt be very pleased to discuss it with you!
If you have the first issue of Sound Practices the circuit is in there too. You don't have the first issue of Sound Practices? Have a look on ebay and US$30 buys a CD of the entire Sound Practices output from Joe Roberts himself.
Basically, the monoblocks have a 6SJ7 pentode driver stage coupled with a 0.1uF Audio Note copper capacitor to the 300B. Blackgates for both cathode bypass caps, and Cerafines in the PSU. Mains transformers and chokes are Hammond, and I was able to save a few bob by reusing the Hammond 1627SE output transformers from the Loftin White.
As the WE91s are monoblocks some method of control is needed. The beauty of 97dB loudspeakers in a typically sized British living room is that they don't need a lot of power, so I built a simple passive preamp to control them. This was the easiest way (rather than an active preamp) and allowed me to control the headphone amp from the same box too.
Starting with a Hammond 12" x 10" x 2" black powder coated chassis the first job was to cover the outside surface with 2" wide masking tape. Experience has taught me that with the best will in the world the metal swarf from all the drilling is likely to mark the chassis unless protected.
Then the layout of the iron and valve sockets can be attached. I drew the layout in AutoCAD first , then printed it out and attached to the chassis along two adjacent sides so I could clear the swarf out periodically. Guess I find it easier this way than marking out on the masking tape.
Then it was a matter of labouriously drilling each hole, starting with a 2.5mm drill bit and opening up with increasing diameter bits 0.5mm at a time. Hole punches were used for the larger diameter holes, such as valve sockets etc. I used a captive power lead rather than an IEC socket. It's much easier to use a circular hole punch for a cable gland than mess about fashioning an odd shaped hole for an IEC socket. Here's the chassis drilled and punched...
And with the masking tape removed...
On to the more interesting bit, building the amplifier. First job, install the iron, valve sockets, tag strips and rest of the hardware. The 1627SE output transformer isn't attached to the chassis yet as I was still using them with the Loftin White...
And inside showing the 300B filament transformer and tag strip...
Then start to install the components and wire up. I sketched out the layout of all the components before I even thought about marking out the chassis layout. This allows me to refine the layout to optimise wiring etc. First lay the components out to see how they actually fit...
Then it's a relatively simple job to build the amp and provided I've got the sketched layout correct I have a master plan to compare my build to. Double and triple checking of the sketch helps!
Finally the amp is complete. Now repeat for the second monoblock...
After the amps were installed into the system controlled by the passive control unit it was time to plug in the valves. The 300Bs were Chelmer badged - they look a little like Valve Art, but who knows what their heritage actually is? Philip supplied some nice NOS Sylvania 6SJ7s, and a pair of new production JJ GZ34 rectifiers to complete the compliment.
Power up, and fortunately no exploding electrolytic caps! Unfortunately one of the GZ34s was arcing. There have been a number of reports on the web about dodgy new production JJ rectifiers and I can now add my own. So I switched the dodgy JJ for the Mullard GZ33 from the Loftin White and we were away.
Straight away I could hear there was more detail and control than the Loftin White, but it wasn't until the amps had been running for a few days that the bass opened up. Now, in addition to the nice midrange and treble there was real weight and extension to the bass. If the WE91 is the equivalent of FM radio the Loftin White seemed like AM radio. I was quite surprised. Not just that the WE91s sounded so good, but that in comparison the Loftin White sounded rather limited in the frequency extremes. So much so that I'd like to rebuild the Loftin White sometime to see just how good I can make it.
The WE91s had taken me a little further in the journey towards musical reproduction realism. I was very pleased.
The easiest place to start is with some valve rolling. Reading on the net suggested that some nice RCA red can 5693 valves would be a decent replacement for the 6SJ7s and some were duly procured through a famous online auction site. Sure enough these were a small but worthwhile improvement. And they look kinda funky in red too.
Weighing up my options for the dodgy GZ34 rectifier I considered a pair of new production Sovteks, but in the end got another Mullard GZ33 from Philip to match the existing one.
Then for the 300Bs. I tried some Electro Harmonix (not the Gold Grid version) first they were a little different but not really any better. Weighing up options, two likely candidates were EH's Gold Grid version (which can be had for under 70 quid a pair) and TJ Meshplates costing a little more. I ummed and ahhed for a while, and after concluding that I could end up spending quite a few quid trying different 300Bs, I took a deep breath and bought some KR 300B XLS from Rod Burman. These are the turbo charged versions and are not cheap, but I really, really like them. These valves have transformed the amps. There is more detail, more control, more tone, more everything.
Monoblock with revised valve contingent - KR 300B XLS, RCA 5693 and Mullard GZ33
The only other change I made was different output transformers. I had ordered a pair of 2.5k c cores from AE Europe in the Netherlands as part of a larger order for another 2A3 amp, and whilst I was building the new amp I broke the c cores in with the WE91s. AE have a good reputation amongst DIYers, both for sound quality and value for money. I could tell as soon as thy arrived they were beautifully made, but how did they sound?
When I first clipleaded the c cores in to the WE91s they sounded very similar to the Hammond 1627SEs they replaced. About a week later I had put an album on and disappeared through to the kitchen when I stopped in my tracks, turned around and went back in to the living room. Where had the bass come from? Deeper and more extended, very nice. Treble is also improved. These are very nice output transformers which also happen to be very good value for money. I had to wait 5 months for them due to as AE were so busy, but I know that one day I'm going to order some amorphous core output transformers from AE. The only question will be silver or copper wired...
I've been listening to these amps for about eighteen months and I really like them. The KRs and AE c cores transform very nice amps to really excellent amps. They aren't particularly cheap to build, probably in the order of 1100 quid, but for those of us not able to design the amps we'd like this is a fine way of achieving really good sound. Just don't try to use them with inefficient speakers.
300Bs seem to need rather a lot of current to drive them properly and I rather suspect that the passive pre I use is not letting the amps achieve their full performance. An active pre amp or another stage may be a worthwhile improvement perhaps. I wonder about an Aikido-style driver between the 6SJ7/5693 and 300B...
And I must try the really nice looking mod that Paul Barker magiced up, adding a voltage regulator tube to the 5693 and doing away with the cathode bypass cap and associated gubbins around the 5693. I haven't plucked up the courage to tweak the amps yet for fear of breaking them. I need another amp so I can have a play without being musicless.
Over the last few months I've been musing why the WE91s didn't sound as good as when I'd first put the KRs in the amps. I reasoned that it could be that my ears had caught up with the performance and I could now hear the deficiencies in the sound. But I felt that the sound had actually deteriorated quite considerably, to a much greater extent than simply me becoming very familiar with the amps. Specifically, the sound had become laid back to the point of me wanting to push the music along. Not slow exactly, but not pacey as I knew the KRs are.
Thinking about things though, it occurred to me that maybe there was potentially a problem with the heater voltages. Philip had specified a simple CRC DC rectification for the 300B heaters as he had some hum when prototyping. The R of 0.1 ohms was set to give the required 5V. So if the KRs draw more current that cooking 300Bs then the heaters may be under voltage. It would have made more sense to have checked the voltage when I first put the KRs, still, better late than never!
4.4V DC. Not good. It seemed very believable that the under voltage has slowly caused the sound to deteriorate. The question is, would upping the voltage reverse this?
Removing the bridge and CRC filter and AC heating the KRs direct from the 5V Hammond filament transformers gave 5.5V AC. I was a bit surprised that the filament transformers should be so much over the required 5V, but regardless of this I plugged the amps into the system and turned on.
After listening to the amps with completely silent DC heating the tiny amount of hum was something of a surprise. Initially the amps sounded quite thin, plenty of "brightly lit midrange", but the bass was very light. But with time the sound evened up and returned to the wonderful sound I knew. Bodging together the 0.1R resistors and a couple of 0.47R resistors I had in stock I was able to get the voltage down to 5.07V. I think it sounded better at 5.5V but I'm trying not to think about that as burning up the KRs prematurely would make them very expensive valves... Still, it's very pleasing to get the sound back to where it was, fortunately valves seem rather tough and tolerant of misuse.
What has been really interesting about playing with the WE91s again is that I have spent quite a bit of time of late playing with 45s, and whilst 45s have a magical top they seem to be a bit vague in the bass. Maybe it's my amp, and I need to play further, but rediscovering the KRs has been really good and spurred me on to tweak further with the VR tube mod and the driver stage. I'd like to try choke input in the power supply too, but that's going to need a couple of rather large mains transformers.