Kef Egg HTS2001

These are good surround speakers but as standard the treble is mushy, and the mid-bass resonates a little too much. The frequency response charts below give a very accurate representation of the sound – a very uneven bass response below 350hz, a ‘small speaker echo’ at 1.5khz in the mid-range and a little too much high end. The driver has a standard 4″ mount so can be easily replaced. The aperture is 96mm and the raised mount edge outer diameter is 102mm. The mounting points are 115mm apart diagonally and 82mm horizontally. These pics are from a Russian site with some more info about the Eggs:


My project:

The driver is SP1498. It has a second order crossover with the drivers connected in reverse polarity, and a zobel on the mid-woofer. The crossover mounts directly onto the woofer terminals. The prominent mid-bass can be tamed with acoustic damping foam, cut to shape, and inserted at both ends of the reflex tube. For my taste, the foam is 2cm deep at the internal input end and 4cm deep at the output end. The mushy treble is fixed by adding MKP1837 or Relcap RTE bypasses on the electrolytic caps.


The tweeter has a 2.2uF series cap and a 0.4mH shunted inductor. The mid-woofer has 0.68mH series inductor with a 12uF shunted cap, and a zobel shunt of 8.2R + 15uF. It’s not unusual and should be relatively easy to upgrade to bi-amping. The issue is the need to change the components – the main inductor for the mid-woofer is a ferrite core and all the caps are electrolytics, most with a +/-20% tolerance. I measured them and only one had the correct value – all the others were at the limit of their tolerance. That’s a huge variation and means the two speakers sound similar but not the same.

So after fitting bypasses, sometime later, I went back and replaced all the electros with Wima poly caps. This had a huge effect on detail retrieval and meant changing the values to keep a balanced sound: the tweeter was too bright so 2R in series fixed this, and that meant  1.7uF series capacitance to match and the same 0.4mH shunted inductor. Q is slightly higher for a better roll off. Due to space, the mid-woofer had to keep its 0.68mH series inductor but with 11uF shunted capacitance, and a zobel shunt of 8.2R + 14.7uF. And unlike the electros, these numbers are accurate 😉

The plan after that was to run them wirelessly from the source with an active subwoofer, and bi-amp them too. (Later, I ditched the wireless due to latency delay).

The bi-amp was very successful. The circuit is a second order like the original, built with 2.4k resistors, sweet sounding Ero MKP 22nF caps (matched to under 1%) and LME49720HA op amps. As it turned out, the resistors values had to change to 1.2k on the high-pass and 4.8k on the low-pass to match the original sound :


The sub amplifier was as cheap as the transformer on its own so I didn’t expect much. And the original amp didn’t take long to blow up so I replaced it with a mosfet “L7”. It sounds great – very clean, smooth, enjoyable tonal quality.


“L7” LJM Design. Output power 150W @ 8R DC +/-56V, 300W @ 4R.
Frequency response range 4HZ -3DB, 350K HZ -3DB
THD <0.01% 100W 1K Hz, Slew Rate = 38V / uS, SNR> 99DB
CFP complementary feedback input stage. Floating Mosfet output.
Class AB. Output stage using IRFP240, IRFP9240, MKP coupling capacitor.

L7 LJM amplifier.JPG


The sub-woofer filter section had only 2 nasty op amps but spaces for 3 to create a phase inverter. So they were replaced with OPA2132 and I added Elna Silmic and ceramic bypass caps on the power supply rails for each op amp. There were no bypasses at all before. I replaced the cheap ceramic caps in the audio path with polypropylene.


Besides a lack of bypasses on the power supply, it didn’t have much of a regulated power supply for the op amps either. It used a simple resistor + 15v zener to drop the 56v rails. So, on each rail, I used two 12v 1 watt zeners to boost this to 24v and used 420r 5w resistor to create a basic pre-regulator before the smallest cheap regulator I could find – a small 317/337 psu. It just fits.

The 317/337 uses low noise 15V zeners at the base to set the output voltage and lower the output impedance and noise. You can read about how that works on this site. It elevates the humble cheap part into something quite decent.

I added the 2nd order crossover for the Eggs so the psu was running 5 op amps and the psu itself: about 60mA on each rail. This is about the limit of the pre-regulator.


The amp thumped at on/off, and needed a speaker protection kit based on the uPC1237 to finish it off. That uses 8mA at startup and 45mA when the relay latches. This is too much to add to the op amp circuit so another circuit is needed to drop the +56V rail to under 25V (the limit of the circuit’s caps). So a 20k 1/2 watt potential divider shunts 2.8mA to ground and the centre tap of the divider is adjusted to feed 34V to the base of a Tip41c NPN pass transistor. 2mA is enough current at the base so the hfe gain & base-emitter voltage drop will enable a 33.3V 80mA output to a 7824 regulator, which in turn feeds the speaker protector circuit. The NPN pass transistor gets hot so it needs a small heatsink but the 7824 stays cool so it runs naked. And free. Ahem. Finally, I changed the connectors so there isn’t a speaker-level input anymore, and the empty holes are used for bi-amp outputs.

And I’m pleased to say the outputs are clean, clear with no audible noise. It looks a mess but in fact works very well.



So with the active sub done, it was time to bi-amp the Eggs. I like Tripath amps so I got some very good value SKN TA2021 amps to try out – just US$20 each with air core inductors, good decoupling, output relays, etc. After burning in for a few days, they sound great but have a little background noise with an ear against the tweeter – as expected. So I didn’t think they needed any modifications at all, and the only annoyance were the tiny speaker terminals, making them tricky to tighten, so I had to use speaker plugs. I made a regulator from an AMB Sigma11 and 80VA toroidal transformer and fitted it into an old ATX psu case. 80VA is the minimum – with the crossover at 3Khz the tweeter will draw about 3W when the woofer is drawing the maximum 20W – it’s a 15/85 split @ 3Khz. So the amps will output 46W and they are >80% efficient so that’s a maximum 60W load on the regulator/toroid. In addition, these are surround speakers so I won’t be using them anywhere near their maximum.


I hooked it up to listen to an original Egg vs the bi-amped and the bi-amped speaker had far too much output over the crossover range no matter whether the tweeter was phase inverted or not, and it sounded like the tweeter output was too much. I replaced the 2.4K resistors in one channel of the active crossover with trimpots and adjusted them until I got a reasonably flat response. I did side-by-side listening with an original Egg and adjusted until the sound was very similar to the original. Then I removed the pots and measured them – the low pass used 5kR and the high pass 1.2KR – so it seems that doubling/halving the calculated values is necessary. I fitted 4.8K and 1.2K resistors and that worked like a charm. I had also dropped the tweeter output with 2R 5W resistor in series and added a zobel on the mid – 9.4uF +8.2R. The zobel caps were later upgraded to Wima MKS 4.7uF with Vishay MKP1837 10nF bypasses. The bi-amped speaker sound is very similar to the original Egg (just a little warmer with less prominent upper mids), and is a little cleaner (less distortion) and more natural (less obvious resonance) so I’m very pleased, and considering it’s cheap to do, I think it’s well worth the effort too.

Last of all, the Eggs got some extra wadding and new speaker terminals. I drilled a 4mm hole in the rear shell to screw on some new terminals. The cable output holes were sealed with Blu-tack. All done and sounding great!

p1070744p1070747p1070746p1070748The sub is a little unusual – I had to fit one under a sofa so I got two 4″ inch subwoofer drivers from Taobao and mounted them on 90-degree, 100mm diameter, pipe bends, at the ends of a 2-metre long pvc pipe – kinda like a push-pull meets transmission line speaker. It was an experiment that worked. 🙂 However, I plan to build a dual 10″ driver subwoofer with isobaric drives to reduce distortion, 1″ thick mdf braced cabinet… put all this gear into a box and run it wirelessly from an APTX-HD bluetooth receiver. I’ve ordered a bluetooth receiver to try with my LG G5 – I’ll do that as a project in another page. 🙂 Thanks for reading!