Radio sound quality - some measurements

[crackle]

Nice work Mark and some interesting results.

Mike

[mhennessy]

Originally Posted by Philips210

It would be good to find out how well the Roberts R600 performs and also the earlier and slightly larger Roberts R700.

Unfortunately, none of my R700s are working yet. I should have an R600 in working condition, but I think that's in the attic, so will have to wait.

In the meantime...

The Roberts R606MB.

Nice sets these, but prone to rubbing voicecoils (it took me 3 or 4 attempts to buy one that was OK). The amplifier is very similar to that used in the Hacker RP25 and related models - the main difference being that the treble control can cut as well as boost. For that reason, I didn't do a "flat" test - that would have meant opening it up and measuring the frequency response of the amplifier in isolation - so I simply adjusted the tone controls for best sound and measured the result.

Overall, it's not bad. It doesn't reach the standard of the R707 - not surprisingly - but it's nice enough. Certainly better than the R505. As is common, there is a large midrange/treble peak (3-6kHz). Or, perhaps it's fairer to say that the treble rolls off prematurely at 6kHz, given how it sits with the 500Hz to 1.5kHz region. I thought the bass could have been better - you can see the peak and rolloff at ~100Hz, but the 100-300Hz region sits quite a bit below where it should be. The expected 400Hz BSC peak is there.

The trouble is, equalisation is expensive. To bring up the 100-300Hz region by 6dB requires the power to be doubled twice. There's only 1 or 2 watts to play with. So that's why the Hacker approach of using speakers that are as large as possible helps - they are naturally more efficient than the smaller ones that Roberts used.

IIRC, the non-MB R606 uses the same speaker as the R707, so perhaps that one will be better?

[Thyristor]

Originally Posted by mhennessy

Unfortunately, none of my R700s are working yet......

Good luck with that one !

I don't currently own an R700, but have owned and repaired several over the years. They are good performers; but a real PITA to work on; IMO better than the R600/R606, but, ofcourse, out-performed by the R707.

I'll have to dig out my Hunter; don't remember which version it is, but its early-ish, ISTR.

Yes, the Sovereign IV is a very good performer, dare I say it; possibly a good compromise of value-engineering; although I'd take a Sovereign III over one any day (or a VHF Herald)

[mhennessy]

Originally Posted by mhennessy

I have other examples of the Aviemore that use different loudspeakers - I'm expecting those to measure rather differently (if I ever find them ).

Well, I've found one of them.

This one uses a nondescript speaker with a tiny magnet. No idea who makes it. We've seen the same unit fitted to the AM-only Ranger...

As before, there are two plots showing both positions of the tone control. Also, another plot comparing this one (black) to the Philips (red). For simplicity, I just used the curves from the tone-up plots. Despite the Philips being arguably more ragged, it's the better sounding speaker as it's more extended. The other one just sounds "middy" because of the lack of HF.

Note how similar the bass is - hard to see any difference at all.

[mhennessy]

Originally Posted by mhennessy

IIRC, the non-MB R606 uses the same speaker as the R707, so perhaps that one will be better?

Found it!

Yes, the R606 uses what appears to be the same LS as the R707.

Both the R606 and R606MB are on the same trace.

As you can see, the R606 (blue) is smoother overall. The mid and treble is pretty good, all the way out to 13kHz (better at the very top than the R707). Note how similar the LF is - clearly the size of the cabinet and the design of the bass lift in the amplifier being common to both.

So, the R606 measures (and sounds) better than the R606MB. Who knew

[mhennessy]

The RP38 with the Fane loudspeaker.

This is the earliest Hunter that I know about (serial number 2208). I really should clean it up and give it some TLC - it's in great condition.

The plot shows the single tone control at both extremes (black and red traces), and the blue trace shows the results after adjustment by ear.

As you can see, this one has excellent bass. The amplifier is different to the later models, and the bass boost (5dB, according to the manual) is fixed in the negative feedback loop. The tone control is a simple treble-cut control prior to the volume control.

Things get quite "interesting" as you go up in frequency. There must be some joke about the Yorkshire Dales here, given that's where Fane are (still) based! Luckily, the peaks and troughs result in a pleasant enough sound, albeit with some of "hardness" because of the 6kHz peak. Extended treble is not on the menu. Continuing to listen to it as I type, I turned the tone control down some more. A notch filter would be a useful addition. Or a tweeter, and a crossover that rolls the main speaker away at 2kHz or thereabouts...

The second image shows the RP38 (blue) vs the RP38A (black) - both with the tone controls set to their "best" settings. Most comparative plots so far have been "normalised" at 1kHz, but the dip at 1kHz prevents that. Instead, I picked 400Hz.

You can see that the earlier model has marginally more bass extension, but the newer model has more level at 100Hz, so most people will think it is the one with "more" bass (it's easy to confuse level with extension, which is just as well for folk who wish to sell small loudspeakers ). Note the box-related similarities in the 200-500Hz region. Above that, things diverge enormously. The Elac is flatter and more extended than the Fane, and avoids having an objectionable peak at 6kHz.

To complete the "Hunter" story, I need to find an example with the Goodmans loudspeaker, and also an RP38 with the Elac (the change happened before the A-model came out). I think I know where these are, but that's for tomorrow!

[mhennessy]

The RP71.

Last one for tonight. There's no beating around the bush - these sets lack treble!

When I restored my first one, I added a capacitor to the feedback loop (in parallel with R24 IIRC) to boost the HF. Sadly, I still haven't rebuilt that one (need to figure out how best to restore the side cheeks), so I can't measure the effect of that here. This example is "stock".

The traces show the tone control at min and max.

The bass extension is fine for a relatively small loudspeaker (6" by 4"). There's bass lift in the feedback network, but a bit more wouldn't go amiss. After 1kHz, the response falls by some 9dB before approximately plateauing until about 5kHz, before dropping like a stone. Listening to it now, it's hard to be sure it's actually an FM set!

Interestingly, the tone control isn't shown on the diagram, suggesting that it was an afterthought. The signal at the DIN socket sounds reasonable at a glance - perhaps a little dull - so it might be possible to slacken off the de-emphasis to help restore some HF. But ultimately, it's almost certainly the wrong loudspeaker for an FM set.

[Radio Wrangler]

Castleford is an awful long walk from the dales. It's down in the (ironically) flatter bit where the coal mines were, so you'd expect deep pits in the response.

David

[stevehertz]

Mark, there's no substitute for objective testing, well done and thanks. One big surprise for me is that the Sovereign 3 that one would have been forgiven for thinking that it should have had one of the best frequency responses, is actually one of the worst! I have a '3' but it is yet to be 'finished' so I've never really studied its tonal signature. On the other hand, one of my favourite portable sounds, the Roberts R707 exhibits a lovely curve for a portable. Yes I agree about the use of its tone controls, but y'know, if one is stupid enough to mis-set them then bad sound will ensue; and if the listener prefers it that way (basically they don't know the difference between good and bad sound), then que sera. It's a self fulfilling prophesy not a fault of the radio, although I do agree there is too much adjustment, but like I say, most people with a good set of ears will just use what is necessary.

Kind of off topic, yet related, I've always thought that a lot of audiophiles sit there judging 'a system' (or an amp, CD player, whatever) and in truth they're actually listening to a pair of speakers. If they weren't so obsessed with using amps with no tone controls then all this nonsense re matching the amp (etc) to the speakers could be avoided by the tiniest adjustment to the relevant tone control to cure say a 'bright' sound or one lacking bass etc. But hey, adding tone controls to a circuit isn't 'pure' is it?

[mhennessy]

I've found my early Hunters, so now we can compare the 3 loudspeakers fitted to the RP38.

Earlier we saw the original Fane loudspeaker. The first attachment here is the Goodmans, and the second attachment shows the Elac. As before, 3 traces: tone at min, tone at max, and tone adjusted for best sound. And as an added bonus, the 3rd image shows all 3 speakers on the same chart, and the 4th image shows the Elac RP38 compared to the same Elac in an RP38A.

Taking the Goodmans first, we can see the same mid-range "hump" starting at 2kHz that we saw with the RP25 earlier. Curiously, it stays up for longer in the RP38 - it's not clear if that's a sample-to-sample difference, or a difference caused by the different amplifier or cabinet design. Whatever the cause, it helps, and I'd say that this Hunter is a better sounding set than the RP25 tested earlier.

As with the Fane, there is the same same nosedive at 10kHz. But note that the Fane is more extended at the bass - an extra 20Hz, which is quite a lot in real terms.

The Elac is nearly as extended at LF as the Fane, and as expected it has much better treble extension too - making it out to 15kHz. In between, there is a dip at 1.5-2kHz (as seen earlier with the RP38A) and a peak at ~4.5kHz (less conspicuous on the A model). That's followed by a dip which was less noticeable on the later model too. Again, without lots more sampling, it's hard to say whether this is sample-to-sample variation, or caused by the differences in the set itself.

When comparing all 3 on the same graph, it's interesting to see the stuff in common (above bass resonance, but below 800Hz) caused by the enclosure, and the differences above that caused by the drive units themselves. At a glance, the Goodmans (blue) appears to be the smoothest, the Fane (red) is the most "mellow", with lots of bass and the broad midrange dip. The Elac (black) might not look all that promising, but in practice it sounds the best because the treble extension helps to balance out the peak at 4.5kHz, and the 1.5kHz dip is fairly benign. The bass is very good as well.

Finally, the Elac in the RP38 (black) and in the RP38A (blue). In practice, they are commendably similar, especially given the different amplifier and cabinet. I'm wondering if the differences at the top end are caused by the grilles - they are very different indeed. Remember that adding a grille with very small holes in it to the T27 tweeter in the LS3/5A actually brought up the HF... A rigorous test would involve removing both loudspeakers and measuring them in isolation - one for the future, perhaps. But to be honest, paper diaphragms were quite inconsistent, which is why the BBC researched plastic alternatives back in the day. These 2 different samples - both nearly 50 years old, in different cabinets, and driven by different amplifiers - are comfortably within 3dB of each other all the way up to 9kHz. That's pretty good IMHO.

[mhennessy]

Hi Steve,

I was rather surprised at the RP72 results - therefore I'm planning to test a few more samples before much longer. I know that I've never warmed to their sound quality, but always felt like I'm in a minority there.

I'm a believer in tone controls - I've got a couple of pro equalisers (parametric and 3rd-octave) in the listening room. Admittedly, those are part of my "R&D" setup, rather than used for routine listening; they help me decide what sort of equalisation needs to be built into the crossover before I try to design it. Oops - crossovers contain tone controls? Hope no-one heard me say that

Of course, measurements are only part of the story, and these aren't intended to be in any way comprehensive - rather, they are just to provide just a bit of insight into what we hear. But for anyone who has say 3 or 4 of the sets tested here, it'll be fun to see if you can correlate what the curves tell you with what you can hear

[stevehertz]

Originally Posted by mhennessy

.....

The Elac is nearly as extended at LF as the Fane, and as expected it has much better treble extension too - making it out to 15kHz. In between, there is a dip at 1.5-2kHz (as seen earlier with the RP38A) and a peak at ~4.5kHz (less conspicuous on the A model). That's followed by a dip which was less noticeable on the later model too. Again, without lots more sampling, it's hard to say whether this is sample-to-sample variation, or caused by the differences in the set itself.

When comparing all 3 on the same graph, it's interesting to see the stuff in common (above bass resonance, but below 800Hz) caused by the enclosure, and the differences above that caused by the drive units themselves. At a glance, the Goodmans (blue) appears to be the smoothest, the Fane (red) is the most "mellow", with lots of bass and the broad midrange dip. The Elac (black) might not look all that promising, but in practice it sounds the best because the treble extension helps to balance out the peak at 4.5kHz, and the 1.5kHz dip is fairly benign. The bass is very good as well.

For me, it also illustrates one of my 'findings' or beliefs from the days when I was involved with designing and building our own hifi speakers (my workmates were the real brains) that a suppressed mid range or lower treble region often gives rise to a nice, 'easy over', natural sound that is devoid of squawkiness, and works very well with the human voice. Hence the Elac is just a 'nice sounding' speaker even though, on the face of it, it has a less than perfect frequency response. Forward mids and lower treble in speakers - to me - are the major cause of listening fatigue; they have a 'sound' that grates eventually.

[Jonster]

That's great research Mark and concurs with my subjective listening of my sets. The first Hacker I owned was a very early RP38 with single tone control and after reading the reviews online I wondered what all the fuss was about, it sounded good but not excellent, great for speech though. I now have a late RP38A which sounds much nicer.

I also have a Sovereign III and an RPC1, the sound is similar, I marginally prefer the Sov for music but the RPC is still very good. I find on both of these the tone controls need to be set to around 2 or 3 o'clock to get the best response.

Another set that I rate is the Roberts R900 which was a successor to the R707 I suppose, with a nice large cabinet. It sounds similar.

[Radio Wrangler]

Originally Posted by mhennessy

Oops - crossovers contain tone controls? Hope no-one heard me say that

Nah, those are bi-wiring simulators. Yeah, that's what they are.

David

[stevehertz]

Bi-wiring? Well, why send the signal to all the drive units via a short 'zero ohms' metal link when you can send it down two lots of (expensive) cable that has a whizzy name?

Bi wiring, two bits of cable, one transferring the treble, the other bass with a resistance of x ohms, hey why not parallel them and get x/2 resistance. They can share the high and low frequency currents.

Back on topic, frequency response isn't the only thing making a good sound, cone resonance can sound awful. Elliptical cones are less prone to this, modern materials are much better so circular cones are just fine. With the older plain paper ones there is quite a difference.

[stevehertz]

Originally Posted by merlinmaxwell

... frequency response isn't the only thing making a good sound, cone resonance can sound awful. Elliptical cones are less prone to this, modern materials are much better so circular cones are just fine. With the older plain paper ones there is quite a difference.

Surely any (cone) resonance manifests itself as an identifiable peak in the frequency response; the two are inextricably linked surely?

[mhennessy]

Resonance implies energy storage, meaning the thing keeps moving after the electrical signal has told it to stop. Luckily, we can test for that: http://www.daytonaudio.com/OmniMicV2/hs17.htm

(that extract from the Omnimic manual saves me having to upload my own. But suffice to say, the 2.5-5kHz peak visible in the large-magent Elac in my VHF Herald that I'm experimenting with now looks pretty messy on the CSD plot. And even if you EQ it down, you can still hear the harshness and congestion buried within. I'll say more if my experimentation amounts to anything publishing).

Oh, and resonance can cause subtraction when the phase is right. The 7kHz dip is caused by that, as the CSD shows.

[Philips210]

Originally Posted by Thyristor

I don't currently own an R700, but have owned and repaired several over the years. They are good performers; but a real PITA to work on; IMO better than the R600/R606, but, ofcourse, out-performed by the R707.

Hi.

I can second that. The R700 must be the worst Roberts radio for servicing. The component leads are squashed flat after insertion into the PCB presumably to keep them in place during the soldering process. This makes it particularly difficult for removing them for replacement or testing. Also, the amount of defective components found during servicing is well above what's reasonably expected. Most of the electrolytics were electrically leaky. A very considerable amount of carbon composition resistors were out of tolerance, so needed replacing. Also the AM IF module is a project in itself. Once removed from the main PCB, it can be dismantled. There were tin whisker problems with the AF117 IF transistors. These were replaced with AF127 transistors for a reliable repair. The IF board also had it's share of out of tolerance resistors.
The FM IF strip also had tin whisker problems with some of the AF116 transistors.This set is a little unusual in that it has separate transistors for the AM/FM IF strips.

The R700 is a nice set that performs well but is very labour intensive to service. I have at least three R700s in my collection but won't be in an immediate hurry to service the others as I've more pressing engagements!

Regards
Symon.

[mhennessy]

Originally Posted by Jonster

Another set that I rate is the Roberts R900 which was a successor to the R707 I suppose, with a nice large cabinet. It sounds similar.

After another deep search of the loft, I found my R900. This is a later one with the conventional "flat" front panel; I've no idea if the "sculpted" models are different.

I had quite a bit of trouble getting this set to sound/measure OK. Like the R707, the tone controls are a bit strange. Also there is a loudness tap, so I had to make sure I hadn't gone past the 12 o'clock position on the volume control.

The bass end was disappointing given the size of the cabinet. Turning up the bass just introduced a hump in the 200-500Hz region, while having a relatively small effect on the real bass. I noted that there is no rear vent with this cabinet, so for fun, I stood the set on one end, and removed the bottom panel - this actually did improve the bass end! It looks like they've picked a drive unit that needs to work in a ported box, or a larger sealed one. Modern drivers intended for mounting in a correctly designed enclosure are very different animals to the high-Q transistor radio speakers of old.

Elsewhere, it's reasonably well behaved. There is a lot of HF output, and I had to be careful with the treble control. The spike at 6.5kHz is unwelcome, but we've seen much worse...

In the same box, I also found my RCS80. This was similarly expensive at the time, but sounds very poor for the money. The bass end isn't terrible apart from the 350Hz bump (like the R900 - probably caused by the cabinet), but the 4kHz spike followed by the rapidly falling away HF wasn't good. To get to this point, both bass and treble had to be turned up pretty much to max.

I've put them both on the same picture (3rd attachment), and the 4th image shows the R900 and R707 together. As you can see, they're quite different beasts. I think I'd pick the bass of the R707 over the extended treble of the R900...

Finally, I thought it would be interesting to look at the R800, which was the flagship model between the R707 and R900. It has one of those plastic Philips speakers that generally sound far better than they have any right to!

I found the best sound was with the tone control slider set to about a 3rd of the way across. The curve shares many similarities with the Hacker Aviemore tested earlier - this shows that the drive unit is the main factor in the way a radio sounds. Both have a large dip at 2kHz, but the R800 manages slightly more bass (perhaps the larger cabinet?) but is less extended at HF. So, not a patch on the preceding R707, but I'm not convinced the R900 was a major step forward in sound quality terms...