SET Friendly MTM, Seas ER18RNX and 27TDFC

I built this in 2007, with a lot of help here from Jay and Zaph particularly with the crossover. I thank these guys most sincerely!

Design was constrained by WAF to about 8.5 inches wide, about 12 inches deep and 1 metre high, also by the amp, my 45 parallel SET giving 3.5 watts (note the decimal point). Don't query the amp unless you've heard a 45 SET, if you have you will understand. Room is 18ft x 14ft.
The speaker is about 42 litres, bass reflex to give good sensitivity.

The Q of the bass driver was calculated allowing for amp Zout, giving Qts 0.51 quite high. This made bass alignment a bit of a problem; the ideal volume for a standard alignment became massive. But it was overcome with use of WINisd. With my volume it was not possible to avoid a hump in the bass, but tuning the port to 30Hz got the hump to below 1dB. Group delay was under 10mS at 40Hz (Zaph's criterion), and if I remember, the peak in group delay was moved to quite a low frequency and was not too massive. In fact this is very much Zaph's 'lean and low' tuning; low is obvious, and lean I take to mean removing or minimising the hump at roughly 80 to 100Hz.

Crossover was very similar to Zaph's one for his MTM with Vifa woofers and 27TDFC, virtually as Jay's suggestion; very simple:
Woofers 1.35mH, 40uF
Tweeter 1.9 ohm series R, 12uF, 0.39mH
The L on the woofers adjusted by ear for BSC, starting value 1.4mH was very good.
Tweeter C was adjusted after SPL measurement to reduce a slight dip in the upper crossover region.

The result?
Bass is very good, adequate quantity, reasonable depth, very good quality; nothing boomy; tight, tuneful, fast. I had expected to want to try different port lengths, or straws or damping, but once I heard it I told myself, do not mess with anything! (I've messed electronics up in the past and never got back to the best).

Overall, sound is tuneful, fine tonality, good rhythms, excellent leading edges (says a critical friend who loves leading edges and dynamics); superb vocals. Highly musical. Just seems to do lots of things right in a way that you don't notice them, they disappear.
I like the combination of amp and speakers a whole lot, and I'd put it this way.
In hifi, you might have decent hifi sound.
Go a level higher, and it sounds like better hifi, you notice the hifi things more.
But at a higher level still, I find it sounds less like hifi; I don't notice the hifi things, they go away, leaving a very natural presentation.
These speakers are getting close to doing that.

Liked a lot by hifi addict friends; a Naim addict dynamics freak says they have some dynamics and rhythms (and he's critical, blunt with it, like Craig Revel Horwood, for my own good, and he knows his stuff).
The system does not sound underpowered in any way; for one thing the amp sounds more powerful than my 300B SETs of twice the power. Definitely all the volume I want, without stress.

Anyway, I can recommend this design to anyone who wants to build speakers for a SET amp and is constrained by WAF.

You will soon understand why I'm posting about these speakers many years after the event.

The system is pretty good.
Teres turntable (acrylic/lead platter), Moerch DP8 arm, Zyx 4D, far better than previous LP12; good valve phono stage.
Marantz CD65 II (TDA1541A dac chip), lots of caps upgrades, NOS converted, lampizated; superb tonality, great sound.
Valve preamp with 27 triode
45 parallel single ended amp, separate PSUs for the two stages, massive PSUs, both 2 pi with chokes, film caps, tuned for fast clean transient response.

  • Like
  • Thank You
Reactions: 2 users
A bit more info.
Sides are curved, laminated from 4mm MDF and plywood, 20mm thick I seem to remember. The curve may reduce some bending modes.
Several braces were fitted including one level with the tweeter, with considerable space around the tweeter.
Tweeters are offset.
Front panel is double thickness, 40mm approx.
The finish is pippy oak veneer.

How to correct Q for amp Zout. Of course this assumes a constant Zout, and factors of the driver being constant, so it won't be 100% but what else can we do, it seems to work out OK.
Correct the Qes for amp Zout Z using driver resistance R:

Q'es = Qes X (Z+R) / R

Find total Q'ts:

Q'ts = (Qms x Q'es) / (Qms + Q'es)

For ER18RNX, 6.3 ohms, with amp Zout 4 ohms:
Q'es = 0.38 x 10.3 / 6.3 = 0.62

Q'ts works out at 0.51, which is a lot different to Qts 0.32

All this tends to explain why some speakers are not good with SET amps; the much reduced damping could mean a loss of control, too much resonance, a totally wrong bass alignment. In my case I had the benefit of analysis with a good tool - and a lot of luck!
  • Like
Reactions: 1 users
Correction to the crossover; I was working from my notes and I had not updated them. On checking them out before doing a parts upgrade I found the following.
The crossover above gave a slight peak and dip in the crossover region when measured. Modelling suggested the following change:

Woofers 1.2mH, 40uF
Tweeter 1.9 ohm, 10uF, 0.39mH
That's what I have and it sounds OK.
  • Like
Reactions: 1 user
Thanks for this sharing.
I made 2 similar projects with 2xER18RNX and 27TDFC first and later with WG300 + DT300.
1. First project in a TL, very interesting because less boxy sound. LR4 slope but I was not satisfied with a second order LC on the tweeter, I used a four order crossover give to my ears a cleaner sound in the treble. The ER18 is amazing in the low midrange. 1.8kHz crossover point
2. A classic bass reflex mtm but the tweeter is in a 6" waveguide. The DT300 is a less fine tweeter than the 27TDFC but more crossover slope possibilities. LR2 and LR4 are possible at 2kHz. Best imaging is with the LR2 slope, a very simple crossover.
I loose the data of these projects...
I measured Zout at the time using signal generator and varying load resistors; worked out as 1.9 ohm on the 4 ohm output. I forget what frequency I used, I didn't do a proper check at different frequencies, but with one resistor load I swept the freq and noticed very little change except where expected, and I forget where the LF and HF started to drop but it wasn't abnormal.