Minimus 7 Crossover Upgrade

This will be a short page.


These are Realistic brand “Minimus 7” miniature speakers with a diecast aluminium enclosure.  I purchased them from a local Tandy store in about 1979 or ’80 as a teenager and have kept them all these years.  They’re of very early Japanese  production and are apparently quite sought after these days.  The tweeter is mounted on a flat steel plate having a passivated zinc coating and painted black on the front.  The first versions apparently had an aluminium tweeter plate and later production was in Malaysia with a plastic plate.  Being pretty basic but ruggedly built, they found their way into the garage where they’ve been used almost daily to listen to FM radio.

A friend mentioned that he was very happy with a “Zilch crossover mod” that he’d done with his own Realistic speakers, so I thought “why not?”

The stock arrangement

Just an inductor across the tweeter and a series capacitor forming a second order high-pass.  Woofers connected directly to the terminals (euch!).

The Zilch circuit

It retains the original 0.4mH inductor, but the other parts are new.  The original BP electrolytic cap was discarded.  I’m no passive crossover expert – that’s for sure.  I don’t even like them, but for low value mini speakers like these it’s the only sensible option.  The new Zobel (R1 & C1) across the woofer flattens its rising impedance as seen by the inductor (L1).  But it does seem odd to have a 1st order LP filter on the woofer crossing with a 2nd order HP filter on the tweeter.  I suspect that the circuit is well tweaked.  Only measurements will tell!  All of the specified resistor values are ridiculous!  I used a standard low tolerance 3R3 for R1, a 39R for R3 and would have used a 2R2 for R2, but 2Ω was actually obtainable.  I just don’t know what some people are thinking sometimes.  Anyway …

The terminal board modified to that circuit

It wasn’t practical to mount the coils at right angles to one another, but they’re spaced apart a bit.

3W metal film resistors¹, low voltage polypropylene caps and a Visaton 1mH air core inductor.  Parts stuck down with double-sided 3M tape and hot melt glue.  Other people use an iron core inductor for lower series resistance, but these turned out to be fine (actually a 2dB loss balances the replacement woofer output very well with the tweeter anyway).  C1 of the woofer Zobel is under the new inductor.

  1.  3W parts are more than sufficient for all three resistors.  The little amp can only deliver about 6V P-P,  so even 1W parts would have been OK.

The resistor (R2) at bottom right is wrong – I had increased the recommended 2Ω to 3R3 (on hand) in an effort to compensate for losses to additional series resistance of the air core inductor, but changed it to 2Ω after taking that photo.

Other (audiophile type) people change the terminals to fancy gold plated binding posts, but the Japanese style spring terminals (not shown) are absolutely fine!

I should mention that my original woofers blew and had been replaced many years ago with “Dia-ichi” general purpose 4 inch drivers, however the tweeters are stock.

This is the repalcement woofer.  It’s very inexpensive, drops straight in and is still available:

Just a pic

That shows new stuffing which replaced the original glass fibre type which was pretty “toxic”.


Earthworks calibrated mic, REW and a Focusrite 2i2 sound card (power amp off camera):

Tweeter alone (after mod):

Pretty spectacular for a cheap 40 year old tweeter, but the crossover itself cannot account for that very steep 36dB/octave roll-off which looks a bit doubtful.  Maybe the mic was too close to the floor.  The next ones were done more carefully:


In the following traces blue is the speaker with the stock filter (modified afterwards) and red is with the Zilch xover.  Microphone not moved and speakers raised up off the floor on a stool (front edge of speaker at front edge of stool), mic at about 800mm.

Just gotta say – even before the red trace was generated, I could tell by the sound of the sweep itself that it was gonna be pretty flat:

By comparison, the blue sweep was audibly very wonky.  And no – a driver was not connected backwards!

These are gated quasi-anechoic measurements with even less smoothing than recommended by Toole.  LF resolution is poor compared to proper anechoic measurements of course.

Horrendous destructive (5kHz) and constructive (9kHz) interference there between the woofer and tweeter in blue.  That’s what you get when you don’t low-pass a woofer!

  • Interesting aside:  I once asked one of those bad sounding ESL panel audiophiles “how are you crossing the panel to the woofer?” His response was “mechanically”.  What?  “I rely on the natural roll-off”.  Well that’s what you get! See above in blue!

By contrast (and as if I even need to say it) the red trace is sensational and the off-axis measurements are good up to 45° (see below under Horizontal polar traces).


Overall about 10% reduction in THD from 500Hz up.

Group delay

Only a minor increase in GD at the LF end as expected from any additional filtering, but much smoother around crossover where it counts.


Needless to say the improvement is incredible for such a trivial effort.

Horizontal polar traces



These were done with the speaker on a stool, fixed mic, speaker pivoted between measurements (something of a poor man’s Toole-style “spinorama”).  No audible resonances – just a slight crossover dip at 3kHz.

And these inexpensive speakers have no audiophile pretentions whatsoever!  Here’s one (to the right →) with audiophile pretentions.  A very well-reviewed (by idiots) American “audiophile” speaker costing over AUD20,000 not only with a deep crossover notch, but an almighty 4kHz resonance to boot! And that’s where hearing is most sensitive.  What a pile of garbage.  A cheap ($50) Realistic speaker with a simple crossover mod (another $50) measures better!

A video just for fun

This is silly really as it’s just a dumb video of the speakers back on the garage shelf playing whatever happened to be on the radio at the time and what you’ll hear will only be mono due to the single mic Android phone, but they do sound good in the garage:

Can somebody please tidy those shelves? 😀

25 thoughts on “Minimus 7 Crossover Upgrade

  1. Very much enjoy your brilliant review and project implementation of the Minimus-7 PZ-2.1 mod. The stock crossover had long been recognized as its weak link. Earlier mods through the years offered minor improvement at best, but fell short of realizing this little speaker’s full potential. Zilch launched his Audiokarma thread inviting community participation to finally exorcise this demon. He began with baseline measurement followed by evaluating all known prior mods as well as ideas submitted by forum members. Pete Basel shared a design (PB-1) that, by chance, he had only recently developed. It was immediately recognized as a winning solution, and then tweaked a bit to become PZ-2.1.
    There was a theme of inclusion in works of the late Zilch that yielded projects accessible to even the most nascent hobbyist. That included relieving the burden of parts sourcing for those with little or no experience. The final component values chosen for PZ-2.1 were partly influenced by availability from a single well known US supplier.

  2. The eBay kit for $39 does not have R2 or R3 at all. In addition, it has R1 and C1 reversed. Did you guys start with this and then change it into what you are showing now?

  3. Very obviously the reversal of the Zobel (R1/C1) doesn’t matter. However, kits sold on eBay will generally be a random circuit put together by some random person trying to profit from you. They often use repurposed/flawed PCBs. The Zilch circuit was not developed here. I merely implemented it and measured the acoustic responses. It is well known circuit and I have shown that it works very well. If the eBay kit omits the L-pad resistors but is otherwise the same, the tweeter will be approximately 2.5dB too bright. The kit is probably useless.

  4. Hi there,

    I’m a beginner when it comes to audio but I’m very curious to do this upgrade. Can yiu pint me i the direction of a parts list. Do you know s good supplier for parts in Canada? I’ve heard you can do this mod without soldering with screwcaps. I’d like to do it this way as I don’t have a soldering iron.

    Hopefully you can help me,

    Thank you

    Jon Welch

  5. I’m in Australia so can’t recommend a Canadian supplier. See if RS or Element14 are in your country. All the parts except the inductor coils are stock standard and should be available from any supplier.

  6. this is one of those proofs they measurements don’t really tell you much.

  7. Perhaps that they don’t tell YOU much. Your standard of “proofs” BTW lies squarely in Audiofoolsville via Chickasaw Falls. Despite that, might I suggest that you put your astonishing hypothesis to Floyd Toole for good measure?

  8. “BANNED” internet troll/audiofool (time-tape: 2 June 2020 at 4:39 pm) returned with a combative diatribe of irrelevance which I threw into a pit of space vipers whereupon it disintegrated magically into cosmic dust. This is not a free platform for ignominious and inarticulate ignoramuses to air their misbegotten, hare-brained BS. There are plenty of other forums where such baloney can be debated ad nauseam. Get your kicks at one of those. 😐

  9. The zilch crossover also measures very well in the Minimus 77 (same tweeter, larger pincushion woofer and cabinet). But it also sounds much better than you might think it would which is really the crux of the matter. Give it a try.

  10. I had a pair of these I had bought in the early 90s. Much like Ian’s pair they are a steel front panel on a dicast frame. Unlike Ian’s mine still have the original drivers. Before Ian had done the Zilch mode on his I had done it to my pair and my ears told me it was well worth the parts and effort. I raved about it to Ian and others at the ESP forums and it turned out ian had a pair as well. So lucky me that he also did the mod too. Lucky in that he has the tools and software to make a reasonable sweep of the results and is more than kind enough to put them up for all to see. Needless to say I am a fan of the mod. In my case is went a bit down the audio fool route. That is to say I bought extras and matched the parts. I used decent 100v film caps, and non inductive resistors. The stock inductors was reused and the woofer inductor was an iron core unit chosen because it was small and had the DCR called out in the Zilch diagram. I spent some time laying all the parts out and hot glued the mess together after assembly. The hot glue was not the ticket and the crossovers came loose. So I again pulled it all apart, peeled all the hot glue off of it all, and the cleaned the parts with acetone. I couldn’t get my crossover compact enough to fit on the input plate.
    I instead assembled the parts together and glued them with epoxy (JB quik) to the back plate. The inputs were pigtailed so I could then solder them to the input terminals. Nothing has seemed to come loose since. Also my values were spot on, likely don’t matter since the drivers are most definitely matched! It made me feel better so sue me. Strike that, Ian used to be a lawyer. LOL.

    These speakers are used for the vacation entertainment system. They are driven with a cheap but more important compact FX Audio D802C Pro Amp/ DAC. It is rated at around 50 Watts into 8 Ohms (more like 40 Watts at reasonable distortion figures) and pairs well indeed with these. Fed with SPDIF from the LG PF1500 LED projector eliminates ground loops from an analog feed. At the head is a Lenovo P320 tiny computer with a Quadro 1000 card. All fits in a single compact box. Literally a movie night in a box. The star though are these little speakers. So small and the sound is clear and clean. With the mod the dialog and instruments are well placed in the sound field now and the speakers just kind of disappear. One isses with small speakers with ports is they have a chuffing problem with low frequencies. No fix for that but a high pass. Hay they are only 100mm drivers so I can cope. Even considering they still get respectably low! Thanks for posting this Ian!

  11. Lawyer? 😯 Mere patent attorney. Still am, but not working any more. Good to hear from you Mike. Bit quiet over at ESP lately. Hope you are managing through this annoying pandemic.

  12. “No fix for that but a high pass”

    Try just plugging the port with a rubber stopper or cork or something and I think you’ll be surprised how much better the bass response is on these speakers. If you’re not planning on cranking them too much and your system has provisions for eq’ing (or your source is a computer) you can add a 12db boost from 35hz, gradually lowering the boost curve up to 100hz and they can be almost unbelievably capable of full range output. Just watch the woofer excursion if you turn it up or lower the boost.

    Don’t even think about doing that if you leave them ported, though! The woofer unloads below the tuning frequency of somewhere in the 100hz area so it has no control below and won’t output any bass anyways below the port’s tuning freq.

  13. I in fact have the ports stuffed for the full length with spun Dacron fiber and this seems to work well. And agreed on the strong bass output in a small room. They have pretty surprising output down low for such small drivers. Tried messing with the EQ a bit and could not make it work well with the limited settings available. Still they do just fine.

  14. So I just got done converting another set of speakers for friends. They bought them off of Ebay and turned out that one of the woofers was burnt and dragging. Rather than return them the seller simply payed for a pair of replacement woofers. The options are slim far as I can tell for the size and the ratings etc. I ended up selecting the Fountek FW100B 4″ Aluminum Cone Midwoofer at parts express. They just drop right in and don’t need to need modification. The Zilch crossover mod was done and they seem to sound pretty good. Not as good as the stock drivers do but I suspect that part of that is that they are new and simply need to be run in. Time will tell. If anyone else has an alternative driver to replace the woofer do tell.

  15. I’ve been pretty COVID bored so I got back into building speakers. I remembered listening to a pair of Minimus 7Ws in my grandpa’s apartment back when I was a kid. I found them in my basement all but forgotten (he passed years ago). I located your excellent write up while doing some research and decide to give it a shot. Here in the US, Parts Express is close to me and I can get good audio-quality components cheap. Reusing the stock coils, the rest of the components only set me back ~$25. I also reused all the wiring, but upgraded the spring clips since Amazon has some cheap drop in binding posts. Textured, outdoor furniture paint and Bondo completed the mod since the fabric covers and boxes were not it great shape. All in all, it cost me about $40 for the pair. Here are some pics of the cross-overs and the finished set.
    It’s tight in there, but I zip tied the components together in separate groups and glued them to opposite walls. This allowed good coil spacing and alignment. I also removed the fiberglass and completely stuffed with Polyfill. I already had some, so no added cost to me.

    They sound great and look good too. I plan to use them as part of a portable 2.1 system with a refurbished computer subwoofer I’m working on.

    Thanks for the great writeup and particularly the measurements so I knew this was worth doing. Pap’s speakers will live on, sounding better than ever.

  16. Thanks Brian. So nice to be appreciated rather than trolled per one of the comments earlier. Cheers.

  17. This article has rather inspired me. I have a few speakers that are crossed ‘mechanically.” It’d be fun to play with adding a HPF to the woofer. Aside from the book by Toole (which I just tracked down through my library) and finding similar speakers with HPF crossovers on their woofer, where else should I be looking for some help on the math to start figuring out what the cross point should be? If I know the slope/cross point for the tweeter, is it as simple as setting a HPF at the same point? Right now I’m considering some Genesis speakers.

  18. If you look at the driver manufacturer’s response graphs you’ll get an idea of where to cross. A general rule of thumb is to cross symmetrically a couple of octaves above the tweeter’s natural roll-off corner, but this depends on the slope chosen and the other diver too.

  19. Thanks for that detailed writeup. I have 7 pairs of 7’s… Ironically…. I changed out the BP for the Audyn 4.7uf audiophile-ish cap from Dayton on a few pair a while back, but didn’t really notice much of a difference. Then again, I didn’t really spend a lot of time comparing them to unmodified ones either. This mod looks intriguing however. I have both Japanese and Malaysian versions, and also have the plastic tweeter on Japanese versions (perhaps replaced prior to them getting to me) and they are definitely not as crisp as the metal plate units.
    Further digging indicates the 1218 woofer is also not identical, despite its part number. Simpler for Tandy/RS I imagine. I suspect they may graph differently, but sound fairly similar however.
    That being said (and on a totally different note) I am partial to the Canton HC100/Plus S/Plus C/Pullman offerings of that era, (which use a slightly larger woofer) and have many pairs of those as well. Perhaps they would benefit from a similar treatment after all these years, but no rush there.
    The question at hand is the volume that the crossover components occupy compared to the single cap/coil. In a larger enclosure, I wouldn’t give it a second thought, but in airspace this compact…. Have you thought to compare internal/external crossover mounting? Not knowing the specifics of the this woofer, it’s guesswork what the ideal volume would be….

  20. When installing the components I had some regard to the volume stolen by them from the tiny enclosure space hence the choice of low power rated resistors, but I wasn’t too worried really. The woofer is a crude thing and if the Qts rose any it could easily be compensated for with tone controls. As they are they can still play the low notes of a piano really well.

  21. In the original audiokarma thread Zilch explored whether bass performance might suffer from the mod’s added volume displacement . He posted a graph overlaying measured responses with crossover inside vs external to the cabinet, and the difference is insignificant. Zilch noted finding more bass response variance between different Minimus-7 samples than between whether or not the mod was installed internally.
    On a different topic: tweeters with plastic faceplates started production in 1987 for all models of Minimus-7. Yet all Minimus-7 were built in Japan until near the end of 1990, when production moved to Korea. Walnut 7W remained in Korea to their end whilst metal cabinet models moved to Malaysia in late ’91, just a year later. Production remained there to the demise of Minimus-7, and through subsequent Optimus and RCA variants that followed.
    The truth is, all design changes of Realistic Minimus-7 took place during Japanese manufacture, with one exception. In1992 Minimus-7W had a model iteration change from 40-2039B to 40-2039C under Korean manufacture.

  22. You poo pooed the thought of a woofer being wired directly to the input terminals (euch!). But one of the most successful designs in history, the Dynaco A-25 did exactly that. The key to it’s success is a Seas 10″ woofer that doesn’t have the characteristic peak before roll off. I still have a pair 45+ years later. I also have a pair of Minimus 7’s that I take on vacation annually. Great write up. Thank you.

  23. Despite their popularity, if the A-25s measured well with such a filter I’d be surprised. Cheers. 🙂

  24. By special order of Space Command who’s come down hard, further commenting on this page must cease at once! It’s a throw-away page detailing a trivial project, but has attracted an inordinate number of off-topic and stupid comments which have required tedious manual deletion. So “Too bad. Too bad”.

Comments are closed.