Bonsai’s X-Altra MC/MM Phono Preamp

Now that I've taken some vacation time, I finally reached the stage for the first power up and scratching my head right now. I'm currently getting the following:

  1. 15V target, 12.28V measured
  2. -15V target, -12.11 measured
  3. 25V target, 13.2V measured
  4. 10V target, 1.01 measured at C46
I've reviewed polarity of the caps and all look correct, including the two that are required to be reversed, C9 and C25. The onboard LED is lit, so currently checking all my solder joints around the PSU.
Minor hickup in the BOM I beleive: item 34 count is 2 but I think should be 6 since R40, R55, R41 and R56 have become 2k2 as well.
On the PCB (underside, May'21 rev.) R46 and R47 are printed as 47R. BOM and schematic are saying 220R, I guess 220R it is but I didn't see this mentioned in this thread so I wanted to make sure.
I solved my issue (for now). Not sure at the moment how this happened, but I received the wrong Amgis transformer, which just happened to fit on the pcb. The one currently on the board has 7.5V parallel secondaries. 15V is what is specified. Fortunately, Digi-Key still has the correct one in stock, which I've ordered. Now to remove and clean up the solder from the board while I wait for the replacement.
As a christmas-bonus, I sucessfully completed my initial test run. Everything looking/sounding good.
I resorted to OPA2187 for the Servo as I couldn't get my hands on anything else. (I thought I had been able to grab the original OPA2188 only to find that I had ordered the VSSOP version - yikes!). Anyway, the OPA2187 still has better offset then the OPA2188 and it also seems to do the job for me.

I am struggeling a bit with determining the MC load/gain based on the instructions in the supplementary file.
I have a Denon DL-103. It's spec'd at 40 Ohms/0.4mV. Some sources say it's 14 Ohms since 2017 but the Manual that came with the Cart says 40.
Output current would be 10 µA

Now in the instructgions it says:

Use the following formulas to determine the required gain
Step 1: RLtot = (.005/Vcart).(Rcart+Rin)
Step2: 1/Rswitch =1/RLtot -1/494

In step 1, what do I put as Rin?
Thanks! WIth everything to off, I get pretty much exactly 1 Vpp output for the 1 kHz @5cm/sec test track on my test LP. So, everything is as it ought to be. I'm still waiting for the correct transformer (I'm using a spare 2x18V that I still had around but that doesn't fit the PCB)

One thing I was interested in with this build is whether I could make out a difference in the sound compared to a tube based preamp. Here's a first test with the same clip from Monk's Criss Cross, DL103 pickup and first through a mostly tube based (E88CC with a SK104 input cascode. Built with kit from and second sample is the X-Altra:
1. TubePre:!Ahn19Qp_SaKQn17VJJkTNI6t8MBk
2. X-Altra:!Ahn19Qp_SaKQpiM7zHTmWwwcTpLa

Recording was made with a behringer UMC204 at 24bit/96kHz and saved to lossless flac to keep the effects of converting to digital as low as possible (given what I have available here)
One thing I was interested in with this build is whether I could make out a difference in the sound compared to a tube based preamp. Here's a first test with the same clip from Monk's Criss Cross, DL103 pickup and first through a mostly tube based (E88CC with a SK104 input cascode. Built with kit from and second sample is the X-Altra:
1. TubePre:!Ahn19Qp_SaKQn17VJJkTNI6t8MBk
2. X-Altra:!Ahn19Qp_SaKQpiM7zHTmWwwcTpLa
Well I put those through my Python spectrum analysis code to show the gross freq response and found significant differences in the bass:

(10dB difference by 30Hz, 20dB by 20Hz apparently) - not clear if its just LPF on the X-altra, or some bass boost in the tube version.

Also I'd assume there's no rumble filtering given the upturn around 10Hz in both spectra. There appears to be slight HF peaking with the valve preamp (less loading capacitance?), and the ultrasonic difference is interesting (though this might be an artifact of different digitization options, or perhaps there's a signal transformer in the valve preamp?).
Across the main part of the audio band they match each other suggesting flat responses 70--7kHz or so. I'd worry about the bass difference - it rather suggests the x-altra version simply has the wrong low-pass time constant with the 20dB difference at 20Hz - perhaps a x10 component error?

[ Ah, I found the schematic for the output HP filter - you have it switched to the 45Hz option I think. Its only a 12dB/octave filter so its not great for rumble suppression, but is killing the subbass alas. Its only 12dB/octave so its not great. ]
Last edited:
There is indeed a transformer at the tube preamp's output. It is optional though as I fitted it to solve a ground loop problem that I encountered. But I may have used it while making the test clip. I made that version a few month ago and I must admit that I don't remember which of those outputs I used. I'll redo the recording with the "direct " out.
The TubePre does not implement a rumble filter and no 20Hz time-constant), So there is no attenuation of any of the low end. Having to power 4 double-triodes, it is also bound to have a bit more noise from the power-supply and transformer at 100/50 Hz so that may have some effect as well.

Regarding the HPF, I have not wired up the switch yet and I had assumed it wasn't engaged when not connected, but I may have misread the scematic there.

There was indeed the transformer enganged in the tube variant. Sorry for the mixup. So here we go again

Without the transformer the high end looks pretty much the same now. What remains is the difference in the low-end with the TubePre not employing any HPF...

Screenshot 2021-12-27 at 16.09.10.png
Last edited:


Paid Member
2003-07-25 10:44 pm
The 20Hz 40 dB per decade slope HPF filter is permanently in circuit in the X-Altra. When you switch the acoustic rumble filter in, it moves the turnover up to 45 Hz 40 dB per decade. If the switch is NOT wired in, the HPF t/o frequency will be 45 Hz.

Those who think 40 dB per decade at 20 Hz isn’t enough should try playing a badly warped record. I have a Manhattan Transfer LP that, without any sort of HPF, is virtually unplayable because the bass cone flap is that bad. With the 20 Hz filter, no flap. I used this as a demo on the Model 1501 preamp at a show where you can switch the 20 Hz HPF in and out. 40 dB per decade is plenty to remove or suppress unwanted LF content.

The 45 Hz t/o frequency HPF (‘acoustic rumble filter’ really is designed for use with classical recordings where this type of noise is evident on some recordings eg Claudio Abbado’s LSO rendition of Stravinsky’s ‘Firebird Suite’.

The RIAA conformance is within +-0.15 dB 20 Hz to 20 kHz - see the measurements section of the article and in the supplementary data on the audioXpress website.

If I understand the measurement recording correctly, it was done with a mic. I would expect the mic preamp noise and room acoustics to dominate the resultant recording. No doubt, the 45 Hz HPF will still be visible in a measurement like this.
Last edited:
If the switch is NOT wired in, the HPF t/o frequency will be 45 Hz.
Dang, now I see it. This was quite a bit of confirmation bias at work when I looked at it and convinced myself that it was at 20 Hz when not wired ;-)
My current speakers fall off quite a bit below 40 Hz so I didn’t really notice it so much during initial listening tests. I‘m working on a fix.
As an additional datapoint I tried comparing the spectrogram with the same clip based on the digital version (Amazon Music HD, 96 kHz, 24 bit) Turns out that version is playing somewhat quicker. I have not seen that before. I've done similar comparisons with other records matching the digital version and what a cart/preamp combination would yield and I've never seen this happen so I'm pretty sure it's not my turntables drive being off (the 1 kHz test tone on the Test LP also still measures in at exactly 1 kHz). The same 1-minute segment is ~500 ms shorter in the digital version.

Anyways, blue line is the digital version:
Screenshot 2021-12-27 at 21.48.10.png

Looks like the digital HD version was mastered quite a bit differently encoding quite a bit more information at the high end - not that my ears are able to pick that up but it's interesting nonethess - albeit a bit offtopic with regards to the X-Altra preamp, so I apologize for that, but I thought it was an intersting find...
Finished wiring everything up - Lo and behold, with all the caps switched into the HPF the curves match up as expected and at 20 Hz, the X-Altra curve is exactly 3 dB below the TubePre's curve. The samples recorded from the record have a bump at 9~10 Hz - I think that would be due to tonearm-cartiridge resonance. When I look up my tonearm, cart and headshell at 9-10 Hz seems about right (compliance for the DL103 should be around 11-12 @10hz and my HS is around 4.5g heavier then the standard headshell for which my tonearm's effictive mass was published so cart+hardware should be at 13g)
I finally got back to finishing up my X-Altra. I had been stalled on the enclosure. Shipping to Canada and the unknown “brokerage fee” that a shipper might apply left me looking for other cases and since I couldn’t find another extruded aluminum case that I knew would fit I made something up out of scraps that I had around. Wood on the ends and metal top bottom and front.

My question is regarding bonding to keep noise to a minimum. It looks like with the aluminum enclosure the enclosure gets tied to the back plane at the pads around the screws on the back plane. Since I have spectate sheets of metal for my enclosure should I use a jumper to tie them all together?

The photos don’t show an aluminum top and bottom plate that each screw into the wood end caps.


  • 62794B8B-58F0-4B64-B77F-618BF4E3E96B.jpeg
    214.8 KB · Views: 45
  • 43C261F1-B848-45A5-928A-762DA56B2930.jpeg
    325.8 KB · Views: 43