Update May 6th: See this new post on what I turned this build in to. This build had too much gain and was too unstable.
Update April 16th: Finishing touches – see the bottom of the page.
Update: I couldn’t leave it alone… I thought I was done, but instead realised it’s better to have the 2nd gain in the usual position between stages 2 and 3, instead of between 3 and 4 as I had it.
This was a very quick change to make – removed one interstage divider, swapped the wiring to G2 around, and adding back in a divider. Before choosing values for the interstage divider I used a 1M pot test the values – came out close to 820k and 220k, very similar to Merlin’s UHG.
Additionally, I added in a bright cap across the G1 pot of the UHG circuit – this brings things very close to the Marshall 2203/04 preamp.
I’ve updated the below to reflect these changes. I’ll re-do the scope traces at some point in the future too – as I guess there may be more changes to make…
After extensive preparations and prototyping, I modified my previous build.
The result is a ~5W amp with the following switching options:
- ‘UHG’ or ‘L2L’ – when in 4 stage mode, this activates the negative feedback around the 2nd stage
- OT negative feedback – select whether the power tube cathode is grounded through the OT or direct to ground as usual
- Selectable bypass options on the 3rd and 4th stages as per the original N5X design
- Tone stack bypass, again as per the N5X
The initial goal was to improve the frequency response of the L2L-style local feedback and make it a bit more versatile.
Along the way I realised it would be nice to have the local negative feedback switchable and, as a result, basically fell in to having a more Super-eXtreme-style build. Whilst testing this, I took inspiration from Merlin’s UHG design to achieve the voicing I was looking for.
A tip I got from the AX84 forum was to use LED biasing of the local NFB (the 2nd) stage to remove a lower frequency pole from the overall response. Trying this improved things (to my ear) and simplified my layout a bit. I also then tried LED biasing on the 1st stage and liked what I heard – I presume I’m getting maximum gain from the stage thus hitting the 2nd stage a bit harder than through partially bypassed cathode biasing.
I borrowed the idea of grounding the power tube cathode through the OT from the Super eXtreme and HO options guide. This is switchable. Also, I added in adjustable bias.
Changes to N5X
This is all achieved by taking the N5X and
- making a new preamp turret board,
- drilling a few extra holes in the chassis,
- coming up with a custom dual gang pot by combining 470k log and 1Meg lin dual gang 16mm alpha pots,
- doing lots of fiddly switch wiring, and
- spending lots of time prototyping or simulating in LTSPICE
Schematic and Layout
Contact me if you want the layout / board plans.
TO BE UPDATED
For scope traces of both the L2L-style and Super eXtreme / UHG modes see this page. Input signal was 1Vpp, 500Hz. Readings were taken at each grid in turn.
TO BE UPDATED
Oscillation / Squeal & Hiss at High Gain Settings
There are two issues:
- With the gains dimed the amp squealed… Between 3 and 4kHz, and it could be triggered by tapping the chassis. Turning down either gain would stop it. The pitch could be controlled by the tone controls and MV.
- It also hisses at an annoying level when the gains are above 80%. I measured ~10mVpp at the 1st grid and ~1.2Vpp at the power tube grid… (The amp is pretty much silent with nothing plugged in.)
For (1), I spent a while chopsticking to no avail. I then used an unconnected oscilloscope probe and noticed that when it was close to the 1st stage grid the squeal would reduce. I thought this was due to the components positioned around that location, but no amount of moving helped or changed the tone. I then tried grounding the centre pin of the socket. No difference.
I then switched to using a socket with a shield for the 1st valve. This seems to have cured the squeal. Although I can’t then look at the lovely valve glow… UPDATE: This turned out to be a microphonic valve – an old Mullard. Using a new JJ or TAD solves this.
For (2), I’m used Vishay ultra low-noise resistors. Seems to help. The amp is quieter than a Blues Deluxe with a Tube Screamer in front of it when at similar gain sounds.
Output Transformer Negative Feedback
I implemented the negative feedback featured in the Super eXtreme and mentioned in the High Octane modifications document. Doing frequency sweeps to generate a bode plot using this feedback on and off (in either mode – L2L or Super eXtreme) I see no difference. And hear no difference as well… Perhaps I need to use the 16 Ohm tap. Or perhaps it’s the position of the capacitor to ground – maybe the bottom of the ‘normal’ Rk and Ck then needs to go to the OT.
I calculate the ‘clean’ output of the amp to be 0.4W before the waveforms start to distort… So not much clean volume at all, which goes along with what I experienced, hence my other build.
The distorted output comes in at 4.35W. Close to the nominal 5W rating.
For there will be many… A couple of issues: a small bit of flubby bass and weird ‘scratchy’ distorted sound behind the note when at higher gains.
Using an EQ pedal and cutting 100Hz solves the flubby-ness. So I may replace the IR LED for the 1st stage bias and go back to a regular Rk and Ck. Likely 2.7k or 1.8k and 0.68u.
The scratchiness could be any number of things; cold joint, too much gain, etc.
I’ve installed the faceplate and also (borrowing an idea from Rob Robinette) a 5″ speaker to the head. This allows me to practice easily. It sounds quite good when used for cleans and light crunches – but heavier stuff becomes a fuzzy mess. Also, as the speaker is not that sensitive, I can have the master volume higher getting some nice power amp distortion.
All in all, I’m very happy with it.