• You can now help support WorldwideDX when you shop on Amazon at no additional cost to you! Simply follow this Shop on Amazon link first and a portion of any purchase is sent to WorldwideDX to help with site costs.
  • Click here to find out how to win free radios from Retevis!

L21 Frequency mod for MB8719

Tolteca

New Member
Mar 14, 2018
31
10
8
33
Mexico
Hello all,

Back at it again.

There seems to be little talk about this modification. This is a strange mod in which you place a variable resistor from L21 to PIN 10 of the MB8719 to achieve a 128 upward channel shift which is amazing compared to the usual 64 upward shift you get from simply grounding pin 10. I have encountered this mod in many of the MB8719 radios I've fixed and I usually scuff at it and remove it without really looking into it and seeing if it's a legitimate mod with good stability. I have encountered the also obscure pin 17 mod in the past which unimpressed me with it's poor stability and tendency to heat the PLL despite producing probably the same 128 upward shift. It is because of this PIN 17 mod and it's terrible stability that I also discriminated the L21 mod. The last radio I encountered with the L21 mod however I did bother to check and besides having that huge upward shift, what impressed me the most was the stability. It seemed to be as good as stock or as good as a pin 10 to ground mod. In this case if memory serves me correctly it was using it's 11.1125 mixer crystal and was achieving a frequency range of 26.175Mhz to 28.045Mhz. With a 11.325mhz crystal it should achieve from 26.815Mhz to 28.685Mhz. In both of these cases we are also using the PIN 11 to ground mod to get a little more range.

I think I'm going to bite the bullet and install this on a radio. I usually go for a crystal switch mod to achieve this kind of range but I realize it has it's own stability issues which in the end may turn out to be worse than the L21 mod. I've managed to make the crystal switch mod fairly stable by standardizing how I lay out my connections. I always use the front mounted switches and hook everything up with solid core wire which I then neatly hot glue to the board. All wire that hangs over the edge of the board towards the switches gets a zip tie. This is to prevent all the leads from moving which is the main cause of drift with this modification. It works quite well but if this L21 mod turns out to be stable I will say bye bye to the crystal switch mod and stick with this new mod that up until now I've been scuffing at. The reason for my reconsideration is my latest attempt at a crystal switch mod which I can't seem to get quite right even though I'm relying on my usual techniques. It's really a clumsy way of doing it now that I think of it and you can only trim the synthesizer for one of the two crystals which in my case is usually the 11.325 because it will be used for upper channels where there is side band. When switching over to the 11.1125 crystal you are on lowers where side band is never heard of and a trimmer cap on the crystal is enough to get you within the ball park.

I'm going to undo the work I had done up to now on this Cobra 148 and go from a crystal switch mod to an L21 mod just to see what it does. From what I can see in the little discussion that can be found of it online the injection of an RF frequency into PIN 10 is somehow tricking one of the logic gates inside.

Here is a youtube video showing the mod.

 

Ok you prolly hate me but...

The Pin 10 is a "tri state" pin.

One state, ground is to the 64 side...
Other State - Pull high - gives you 128
Float - don't care

Do NOT exceed Pin 9 voltage EVER - that the purpose of the limitation or resistor value - keeps Pin 10 from drawing excessive current and exceeding Pin 9 - you can run up to 8 volts, but just don't make pin 10 higher than pin 9 - then you'll get a dead chip in the deal.
 
  • Like
Reactions: NZ8N and kaos513
So I just realized that this is pretty much the same as the pin 17 mod except that you are DC isolated from PIN17 through C87. This is the 34Mhz trippler signal that you're feeding into pin 10. So apparently what is probably happening is that you are making the gate switch so quickly between high and low that it ends up getting confused. There is also the need for the right DC voltage which is about half of PIN 9.

I wonder if it may be wiser to couple the 34Mhz into pin 10 through a capacitor and add the DC component with a voltage divider resistor or trimmer pot instead of hooking it directly from L21. Could this make it better or make it worse? Anyone tried this? I'm thinking something like a 100pf capacitor maybe? Perhaps I should just go with tried and true.
 
NO!

Pin 10 is part of the input programming...

It's a "tri-State" means that if the ROM mask allows, it is used. Else any application of the pin 10 High or Low - may not work because the ROM invalidates the pin as a reference for Divider...

So if you want that pin to work - you have to check and see if the 64 or 128 divider access is working on it.

You cannot feed RF of any kind into it.

TR20 - to L21 - think this thru.

R138 is the LOAD resistor for TR20 to send power to - If TR20 is "resonating" and providing a <2MHz "Clock" to Pin 17 - C111 filters out our ~1MHz signal and leaves the result of TR 20's "lock" from the TP10, and Pin 10 UHIC chip "35MHz" Beat Frequency as more of a DC component bias from TR20 working in the loop.

So if things are good, the "chop" of 1MHz won't be hard on Pin 10 - it will see it more like a floating voltage that TR20 sends thru L21 - into R138 C111 combo. Its' also the WHY you have to use a dropping resistor - to shield pin 10 from excessive RF - you want a DC component - not the RF. You are making it do a "carry" option. - as long as the ROM Mask allows this so don't try to make it flip-flop too much.

If it's grounded it divides by 64 - if it's enabled (voltage present - read the notes on the vid) will divide by 128.

So you will need to do the Xtal conversion - from 11325 to 11.125 to "open up" what the MB3756 can do because you now can divide down even LOWER means you can change your channel plan to Low-Med-High. All by floating grounding or raising pin 10....(Again if ROM Mask will allow)

Schematically areas involved...
Cobra148L21R138C111.gif

The Story...
Process Page 1
Cobra148L21R138C111P.gif

Process Page 2
Cobra148L21R138C111P0.gif

Process Page 3
Cobra148L21R138C111Ps.gif

Process Page 4...
Cobra148L21R138C111Ps1.gif

Process Page 5...
Cobra148L21R138C111Psd.gif

Process Page 6...

Cobra148L21R138C111P2.gif

Process Page 7...

Cobra148L21R138C111P1.gif Process Page 8...
Cobra148L21R138C111P3.gif

There ya' go...

Hope this helps!
:+> Andy <+:
 
Or, you take an LM7805 regulator, dial it down to +4V and feed it into pin 10 of the MB8719 PLL and get the same results. Been doing that for years to get the expanded "192 Channels" on the MB8719.

I just install a 1N4001 diode inline with the +5V pin on the LM8705 regulator so it drops down to about +4.2-4.3V. It's a fully isolated voltage so you're not trying to pull a voltage from another spot on the PCB. You can use a variable trimmer pot to dial it in if you like. I just do the quick 'n' dirty method with the diode and it works well.


~Cheers~
 
  • Like
Reactions: Dmans
Huh. I'd read a Brazilian made MB8719 datasheet that talked about this mid state but I'd never seen it actually used before. I did think this had something to do with this mod since I was always reading about 4V on pin 10 when this mod was activated. I just thought it was so weird where they were sourcing the 4V from that it must have had something to do with the RF present on that line.

As I understand pin 10 is normally high through an internal pull up resistor, instead of a 7805 or something like that can I just use a resistor from pin 10 to ground so that it along with the internal pull up resistor creates a voltage divider that will give me 4V?

I tried doing the mod the normal way and it did not work as expected. What it did do is act as if PIN 10 was grounded giving me the normal channel range with the 11.1125 crystal, just no higher channels. I just realized that this radio has the RCI8719 chip. It isn't the -99 version of it so PIN 10 is usable, as I understand the -99 has this pin permanently high and if you try to ground it the chip pops. On this RCI8719 however this pin can be grounded or left disconnected with the appropriate effect, but I do not know if it will have a "mid" state.
 
The mid state voltage is dependent on the silicon.
The variation is +/- a tenth of a volt or so.
It can be touchy on some chips. A few unstable, but resolved by swapping the chip.

It should NEVER be connected to Fin, which a lot of people do and this causes other issues with the chip.

Otherwise, yeah it works.
It's not intended to work this way, but it does.
When you do it you are sort of abusing a silicon bug.
 
  • Like
Reactions: ExitThirteen
Martian is correct. I have had 100% success rate with the original Fujitsu MB8719 chips. I have not experimented with the RCI variants, however. Most of the Fujitsu chips tend to work between 4.2-4.3V, altho I have seen some work at 4-4.1V.

And, he's also right that you shouldn't connect it to Fin, the voltage to pin 10 should be a regulated, isolated DC source. Hence the use of an LM7805 regulator IC.


~Cheers~
 
I tried doing it with a variable resistor from pin 10 to ground with no results. At around 4.5v the chip just switches from low frequencies to normal frequencies with a bit of a grey zone in between where the chip can't decide what to do. I tried really fine tuning it in but I can't seem to get it work. This mod is probably best done with the original MB8719 chips. Not sure if this is the case with all RCI8719s though.

I have another 148 with an actual MB8719 that I will try this on.
 
It floats - "Don't Care" - uses either NPN (Grounded) or PNP (High) - then it turns on and selects Divider 64 (for Ground ) and Divider 128 (Pulled High)

Just current has to flow - and hence the "tri-state" - stays off with no connect and remains "Don't care" so as you approach the internal reference - that is the "float" state level.
 
I have a few of the RCI variants and the one that I can remember trying this did not work.
Not to say it doesn't work because a sample size of one is hardly definitive.
There are also a couple variants of the RCI variant as I recall as well.
You can get this working with a voltage divider. Make one of the resistors variable.
It can be really touchy so a cheap 10 turn works best to dial it in.
 
I did use a variable resistor in combination with a fixed resistor to get the best selectivity within the 4.4 to 4.7 volt range which is where the transition takes place. Above or below that gives me either low or normal frequencies. Can't seem to dial it in and make it do the thing so the bug doesn't seem to be a present exploit on this one. Would not discard all RCI8719's though, next time I encounter one I will try again.

I'll finish this one up since I'd already almost finished my crystal switch mod and try doing this with the other 148.
 
It was probably the RCI variant of the PLL that doesn't work.
I would use an original Fujitsu and try it again.
 

dxChat
Help Users
  • No one is chatting at the moment.
  • @ kopcicle:
    If you know you know. Anyone have Sam's current #? He hasn't been on since Oct 1st. Someone let him know I'm looking.
  • dxBot:
    535A has left the room.
  • @ AmericanEagle575:
    Just wanted to say Good Morning to all my Fellow WDX members out there!!!!!