|
Vintage Radio (domestic) Domestic vintage radio (wireless) receivers only. |
|
Thread Tools |
24th May 2017, 9:19 pm | #21 |
Dekatron
Join Date: Feb 2007
Location: Leominster, Herefordshire, UK.
Posts: 16,528
|
Re: Improving the frequency response on a Roberts RM33
OK, it's a bit late now, but I'll have a play on what you say tomorrow. It sounds like getting rid of the middle is perhaps unhelpful and I didn't boost bass enough!
The 100n caps into and out of the volume control have a noticeable bass limiting effect- increasing both of them to 470n or so wouldn't hurt nor increasing the speaker cap. 470uF would hurt there. If the speaker is just too small, there may not be much mileage to be had out of bass lift, but it's worth a try.
__________________
....__________ ....|____||__|__\_____ .=.| _---\__|__|_---_|. .........O..Chris....O |
25th May 2017, 10:49 am | #22 |
Dekatron
Join Date: Sep 2011
Location: Rugeley, Staffordshire, UK.
Posts: 8,809
|
Re: Improving the frequency response on a Roberts RM33
Thanks Chris. I'll give it a try.
__________________
A digital radio is the latest thing, but a vintage wireless is forever.. |
25th May 2017, 12:01 pm | #23 |
Dekatron
Join Date: Jul 2007
Location: Evesham, Worcestershire, UK.
Posts: 4,241
|
Re: Improving the frequency response on a Roberts RM33
The good news: I've finally found my RM33! I'm really glad about that, as I was feeling very uncomfortable with not knowing where it was. I might have a lot of radios, but I have detailed documentation about where they came from and what condition they're in, and what I've done to them, and know where they all are. In fairness, this one was well-hidden in the bottom of a box that I'd already looked in, but even so, I'm happy now. But, enough of my problems
The bad news: it's the other (earlier?) version with the TDA5700 and single-transistor preamp. But, in the same box was my RM30, which does have the TDA1071. From a quick glance, the loudspeakers look identical, and if you ignore the FM presets on the RM33, the electronics is the same. So any tinkering with the RM30 should be equally applicable to the RM33, but I will check both carefully to validate that assumption. I didn't have time to measure them this morning - I literally uncovered them on the way out when I decided to have a quick second look in the back attic - but will take a look tonight. As to the capacitors, I agree that 150uF would seem marginal at first glance, but we really ought to take 30 seconds to think it through before jumping to conclusions. After all, we can be quite confident that the designers at Roberts knew about 1/(2piRC) The loudspeaker is a nominal 12 ohms. If we assume the loudspeaker is a resistive load (it isn't!), then that gives an electrical -3dB point of 88Hz. Precise calculations are impossible until the impedance curve of the loudspeaker has been measured, so we'll have to accept that this is the best we can say at this point. Remembering that the impedance will shoot up at resonance, which won't be a million miles away from 88Hz, the reality might well be better than predicted - indeed, there might even be a useful resonant peak caused by the small cap. But let's wait and see... Meanwhile, given the size and type of loudspeaker, I'll be amazed if it can give much if any output below 100Hz - therefore 150uF doesn't actually alarm me at this stage. It just looks like they deliberately chose a small output capacitor to save putting LF into a speaker that won't actually reproduce it. That in turn saves energy, permits a smaller heat sink on the output IC, and potentially extends the life of the drive unit. In short, good engineering (although it would have been even better to set the LF roll-off before the power amplifier to save it clipping early). For an example of where I did the same, this simple project includes a 2nd-order (12dB/oct) high-pass filter at 100Hz. Thinking about the 100n cap after the volume control, if the Zin of the TDA2611 is 47k, that results in a pole at 34Hz or less, depending on the position of the volume control. The one in advance of the volume control is harder to calculate precisely, but it'll be very similar. So, there's really no point changing these - their LF point is well below that formed by the output capacitor, and even with that increased, how much 34Hz will that loudspeaker give us? But ultimately, let us not confuse LF extension with LF level. What Steve needs - and having listened very briefly this morning, I'd agree - is bass boost. A good 10dB or more. Whether that's feasible really depends on what the speaker can give, and I predict that decreasing one of those 100n caps might be on the menu to prevent the amp from overloading. More later, Mark |
25th May 2017, 12:40 pm | #24 |
Dekatron
Join Date: Sep 2011
Location: Rugeley, Staffordshire, UK.
Posts: 8,809
|
Re: Improving the frequency response on a Roberts RM33
Thanks Mark. I look forwards to your update.
__________________
A digital radio is the latest thing, but a vintage wireless is forever.. |
25th May 2017, 4:12 pm | #25 |
Dekatron
Join Date: Sep 2011
Location: Rugeley, Staffordshire, UK.
Posts: 8,809
|
Re: Improving the frequency response on a Roberts RM33
Just out of interest, access to the print side of the components associated with these mods is nigh on impossible unless a bendy, multi-point, 'ribbon' PCB is removed. Opting not to do that, I snipped the three caps off as far above the main PCB as I could and then soldered the new components to their original leads. The challenge being to not overcook the soldering and causing the cap leads to come free from the PCB!
__________________
A digital radio is the latest thing, but a vintage wireless is forever.. |
25th May 2017, 6:00 pm | #26 |
Heptode
Join Date: Apr 2008
Location: Eastbourne, East Sussex, UK.
Posts: 671
|
Re: Improving the frequency response on a Roberts RM33
It might be interesting to try the bass EQ circuit Grundig used in the 210 Yacht Boy as was being discussed in another thread. Are the speakers of a similar size?
|
25th May 2017, 6:38 pm | #27 |
Hexode
Join Date: Jun 2003
Location: Burntwood, Staffordshire, UK or Kabaty in Warsaw Poland.
Posts: 438
|
Re: Improving the frequency response on a Roberts RM33
A few years ago I experimented with a one transistor baxendall network built on to a piece of vero, using presets as bass and treble controls, lashed up into one of my RM33s. It may have been the very one you had off me Steve !
Results were very good, sound quality vastly improved, so its well worth doing. I only lost interest because due to the construction of the RM33 as Steve mentions, it was difficult to incorporate the pcb in the set. But Im sure its possible with a little thought. Mark |
25th May 2017, 7:21 pm | #28 |
Dekatron
Join Date: Sep 2011
Location: Rugeley, Staffordshire, UK.
Posts: 8,809
|
Re: Improving the frequency response on a Roberts RM33
Sounds good, can you let me have the circuit Mark? or better still, post it on here. Cheers.
__________________
A digital radio is the latest thing, but a vintage wireless is forever.. |
25th May 2017, 8:37 pm | #29 |
Hexode
Join Date: Jun 2003
Location: Burntwood, Staffordshire, UK or Kabaty in Warsaw Poland.
Posts: 438
|
Re: Improving the frequency response on a Roberts RM33
I think this was the circuit from an old book of Tandy transistor circuits. Can't remember if I inserted it before or after the volume control, but it didn't work in one of the positions. Obviously you won't need VR1 and disregard the pencil scrawl. I have used similar circuits in various sets over the years.
|
25th May 2017, 8:52 pm | #30 |
Dekatron
Join Date: Jul 2007
Location: Evesham, Worcestershire, UK.
Posts: 4,241
|
Re: Improving the frequency response on a Roberts RM33
Some initial measurements...
The first plot shows my RM30 with the TDA1071 (black) compared to my RM33 with the TDA5700 (red). As you can see, they are quite different above 1kHz. The TDA5700 set is quite a bit brighter. Is this because of the electrical differences, or the speaker/cabinet? So the next plot shows the results of the loudspeakers driven directly from my test amplifier. As we can see, they are close enough - given these are cheap radio loudspeakers that are 35 years old, the differences are pretty minor really. So, we can conclude that the difference between the sets are caused by the electronics, not the loudspeakers or cabinets. As noted above, Steve's set is electrically the same as my RM30. So all further investigation will be done on my RM30. For people with the TDA5700 and the single transistor pre-amp, a different solution will be required. Having established that, let's comment on the sound quality. As can be seen, they ain't great! Note how rapidly the bass and treble falls away below 200Hz and above 6/5kHz. Improving this will be a non-trivial exercise. Swapping speakers, as suggested earlier, won't be easy because it's an odd size (136 by 70mm, according to the manual, which is 5.25 by 2.72 inches). Incidentally, bridging the 150uF output capacitor made practically no difference. Perhaps a dB at best, but that's within experimental error, frankly. I'm really not surprised, for all the reasons I detailed earlier. I measured the resonant frequency of the drive units as 182Hz (RM30) and 189Hz (RM33), which tallies nicely with the curves. Superimposing the "end to end" plot and the raw loudspeaker plot showed that there was almost no difference below 1kHz - perhaps a dB or 2 on occasion, and the main differences were in the treble region (6dB at 10kHz), which shows that while the receiver electronics are pretty flat in the LF region, there is a distinct HF cut in designed in - over and above the usual de-emphasis. So what can we hope to achieve? Realistically, when you look at how rapidly the output falls below 200Hz, I doubt we'll be able to flatten that out. Between 200 and 100Hz the slope is 24dB. Each 3dB is a doubling of power, and that's 7 lots of 3dB! Using a 3rd-octave equaliser between the CD player and the SMG40, I experimented until it sounded reasonable. I kept it as simple as possible, to give us a chance of implementing it. See the before-and-after plot, and the photo. You might wonder why the 10kHz slider isn't all the way up. Honestly, it hardly made any difference to the sound - the drive unit is so far down at that frequency that it's hardly worth trying. Even with my best efforts, it hardly sounds any better than a small AM radio. That peak at 6kHz disguises the lack of real treble, providing you're only using it for background. As for implementing those curves as a modification to the set, I'll give it some thought, but the data presented here should be enough for anyone else who wants to have a go. Hope this helps, Mark |
25th May 2017, 9:10 pm | #31 |
Rest in Peace
Join Date: Jul 2011
Location: Bridgnorth, Shropshire, UK.
Posts: 787
|
Re: Improving the frequency response on a Roberts RM33
If bass and treble boost are desired, how about suppressing mid frequencies?
Put a tuned circuit in series with the speaker such that it blocks at around 1kHz. Lousy Q will ensure it's not a deep notch, just a dip in the response. If necessary, unpolarized electrolytic of the sort used to couple to line yokes can be used. Just turn up volume for bass and treble boost! On a similar topic, I once bought a DAB alarm clock with a tiny speaker at a car boot. I wired it to a nice big speaker thinking I'd get some phat bass. Alas no - it still sounded tinny. It seems the bass was filtered out in the radio. I looked into trying to defeat this filtering but it would involve I2C comms interception. That seemed too much like hard work. |
25th May 2017, 9:15 pm | #32 | |
Dekatron
Join Date: Feb 2007
Location: Leominster, Herefordshire, UK.
Posts: 16,528
|
Re: Improving the frequency response on a Roberts RM33
Quote:
The lack of any response from the speaker below about 200Hz seems to be the main problem. I suspect that getting any usable output below about 150Hz is a non starter even if the normal listening level is 10dB or so below max welly.
__________________
....__________ ....|____||__|__\_____ .=.| _---\__|__|_---_|. .........O..Chris....O Last edited by Herald1360; 25th May 2017 at 9:24 pm. |
|
25th May 2017, 10:14 pm | #33 |
Dekatron
Join Date: Jul 2007
Location: Evesham, Worcestershire, UK.
Posts: 4,241
|
Re: Improving the frequency response on a Roberts RM33
Yes, loss networks do indeed dip the midrange.
I have a tentative solution. See the attached, which shows the original (black), the EQ'd version from my last post (red), and my first stab at a mod (blue). All are normalised at 1kHz. It's a big improvement on the original, and sounds quite OK over a decent range of programme material, but it does cause a loss in available maximum volume. It should be OK for a bedside set, however. At maximum, it's still quite loud with compressed stations like Radio 2 and Classic FM. Even Radio 4 is nice and clear. But having to turn the volume control further round is the price to pay for simplicity (though it might be possible to get more gain from the power amp - I'll take a look at the datasheet in a moment). The mod involves replacing the 22k resistor (R33) for 100k, and replacing the 4n7 capacitor (C34) for 33n and 10k in series. That's it The 4n7 was killing the treble (such that it is!), so removing that makes things a lot more lively. The new network that replaces it provides bass boost instead of treble cut. At the frequencies where the 33n is effectively a short circuit, the 100k and 10k resistor work as a decent attenuator. But at low frequencies, the 33n starts to tend towards an open-circuit, so the bass goes up. It's a bit lively in the 1.6 and 2.6kHz regions, but it sounds OK on the whole. In a typical living room or bedroom, it'll be fine. Much better than before. I emphasise that this is my first iteration, arrived at in about 5 minutes, so no-doubt there are better (and much more complex) solutions out there. But I'm quite happy with this, given how much I'm getting paid for all my efforts Cheers, Mark |
27th May 2017, 12:50 pm | #34 | |
Dekatron
Join Date: Jul 2007
Location: Evesham, Worcestershire, UK.
Posts: 4,241
|
Re: Improving the frequency response on a Roberts RM33
I'm a little surprised that no-one has commented since my last post. After all, it offers a pretty reasonable answer to this:
Quote:
It was satisfying to distil this down to such a simple solution, and I seriously doubt it could be improved without significant extra complexity (involving extra transistors or op-amps or similar). When I get some spare time, I'll investigate a similar fix for the TDA5700-based versions, and I'll also take a look at the RCM1, which has a very similar audio section. Mechanically, I found it fairly easy to work on the set. With 2 screws undone, you can hinge forward the PCB from the chassis sufficiently to solder properly. OK, it's not a Hacker, but also it's not an R707 Like my recent work on the Grundig N210 Yacht Boy, this simple (and reversible) mod turns these sets from mere display items into radios I can actually use. I hope that others feel the same. All the best, Mark |
|
27th May 2017, 2:20 pm | #35 |
Dekatron
Join Date: Sep 2011
Location: Rugeley, Staffordshire, UK.
Posts: 8,809
|
Re: Improving the frequency response on a Roberts RM33
Mark, seems like your last post and my new one (this one) almost crossed 'in the post' as this morning I have performed the mods you suggested. Forgive me, I don't work on my sets every day, but today I had an opportunity.
The sound is much better, more than acceptable in fact given the size of the speaker. Whereas before there was virtually no bass instruments to be heard, now you can. Similarly, at the other end of the scale I can now hear cymbal crashes and cowbells etc. From the mid range upwards the balance is quite probably as good as it is possible to attain. The bottom end could possibly benefit from a dB or two of lift if possible, but really, this mod was so simple to perform, I am quite happy with it as I have already said. Certainly, if boosting the bass a tad more required more complicated component swapping etc, I'd rather stay as we are. Thanks Mark (and Chris) you've been a great help!
__________________
A digital radio is the latest thing, but a vintage wireless is forever.. |
27th May 2017, 3:04 pm | #36 |
Dekatron
Join Date: Jul 2007
Location: Evesham, Worcestershire, UK.
Posts: 4,241
|
Re: Improving the frequency response on a Roberts RM33
Glad to hear you saw it, and glad to hear that it's working for you
If you bring the bass up any more - even just by a small amount - then it gets really muddy. Knowing your tastes, I tried, believe me Actually, it's flatter than many of the radios I measured recently. What you're hearing now is a lack of extension, and that requires a new loudspeaker, sadly... |
27th May 2017, 3:32 pm | #37 |
Dekatron
Join Date: Sep 2011
Location: Rugeley, Staffordshire, UK.
Posts: 8,809
|
Re: Improving the frequency response on a Roberts RM33
Fantastic, I'll put it back together now then!
Thanks again Mark. Like you say, the mod 'converts' a cute, collectable radio into one that is much more useable and listenable to on a daily basis. Mine will sit above my (computer) work desk within arm's reach! Out of interest, a previous owner (not my best mate Pilot Mariner who I had it off) had twiddled all of the internal tuning components. As I reported, the varicap voltages had to be reset, but only recently having tried it out on LW and MW, they too were twiddled to hell. I did my normal thing of interpreting the alignment instructions to be used wrt actual stations at each end of the scale(s), and it is now 'dial accurate' on LW, MW and VHF.
__________________
A digital radio is the latest thing, but a vintage wireless is forever.. |
27th May 2017, 3:53 pm | #38 |
Dekatron
Join Date: Jul 2007
Location: Evesham, Worcestershire, UK.
Posts: 4,241
|
Re: Improving the frequency response on a Roberts RM33
And rather splendid in walnut too. In the flesh, these sets are somewhat smaller than most imagine, which just adds to the cuteness. I have an earlier RM50, which is quite a bit larger, and rather ungainly as a result. That comes with a 4 by 6 inch speaker, but could also use some "help" in the EQ department. One day...
Oh, what's the serial number of yours? I've been assuming that yours is newer than mine - perhaps influenced by comments made about the walnut version on the Roberts Yahoo group - but I think I'm wrong. I've just noticed that my RM33 with the TDA5700 uses plain orange capacitors, which came after the earlier "tropical fish" types. Also, my RM33 uses TO92 transistors rather than Lockfits. I've found a date code of 1977 in my RM30, but nothing I'm confident about in the RM33. But it's looking like the TDA1071 design is actually the earlier version. Of course, your case could be newer than the innards Anyway, my TDA5700-equipped RM33 is 522727, and for what it's worth (probably not much in this context), my TDA1071-equipped RM30 is 306361. Cheers, Mark |
29th May 2017, 4:05 pm | #39 |
Dekatron
Join Date: Sep 2011
Location: Rugeley, Staffordshire, UK.
Posts: 8,809
|
Re: Improving the frequency response on a Roberts RM33
Just to finish off, I know people like photos, here's a couple of the set all back together and polished up, but most of all, sounding great now! Thanks to Mark and Chris again.
BTW, the pound coin resting on top gives an idea of just how small and 'dinky' this set it.
__________________
A digital radio is the latest thing, but a vintage wireless is forever.. |
30th May 2017, 12:26 am | #40 |
Dekatron
Join Date: Jul 2007
Location: Evesham, Worcestershire, UK.
Posts: 4,241
|
Re: Improving the frequency response on a Roberts RM33
For anyone who has the later* version with the TDA5700, I've done the same job on that one...
*Since post #38, I've noticed that the service manual that covers the TDA5700 version says "From serial 515006" on the first page. So that confirms it. The TDA1071 was used in the earlier versions. This one uses a transistor preamp (TR3), and we'll add 3 components to it. These needs to be mounted on the rear of the PCB, so some dismantling is required. It's much easier if you undo the 2 screws that secure the 3 wavechange switches and allow those to come away with the PCB. And the RM33 is easier than the RM30 First, we add 1k in series with 220n between collector of TR3 and ground. This forms a loss network in conjunction with the output impedance of TR3 (4k7). At lower frequencies, the 220n tends towards an open-circuit, so the bass is able to rise in level. Next, to bring the treble up, we add a 150n capacitor across the emitter resistance. This value is reasonably critical - if you go up to 220n, it's a bit "squawky" in the midrange, but if you pick 100n, there isn't quite enough lift. But obviously, feel free to experiment to taste. With the TDA1071 version, I got the HF lift by removing the (over-cooked) de-emphasis. Here, I decided to leave it in place, as the transistor gave me the ability to HF boost directly. I did check it, and found that it's actually spot-on with this set. OK, that's sort-of academic, given that there's no "tape out" or similar, but it felt better to leave it alone. Attached are the before (black) and after (blue) plots, and the schematic showing the additional components. As before, the goal was simplicity. I'm pretty happy with just 3 components tacked on the rear of the PCB - that can be readily removed if desired. Hope this is of interest, Mark |