QG8 internal regulation

cancel
Showing results for 
Show  only  | Search instead for 
Did you mean: 

QG8 internal regulation

4,414 Views
admin
Specialist II
Hi guys,
I am planning to run the QG8 at 3.5V but my power supply is 5V, so far I am using a diode and a resistor to drop the voltage down to 3.5V and I am getting less than 50mV of noise P-P at the VDD of the QG8. Does anyone know what is the acceptable noise level at VDD?

Thanks in advance.
Labels (1)
0 Kudos
19 Replies

1,175 Views
baddad
Contributor I
I don't see a recommended spec for noise, but I would make sure you put the recommended 10uF bulk and 0.1uF high freq capacitors on Vdd.

Make sure you put them as close to the pins as possible.

BadDad
0 Kudos

1,178 Views
admin
Specialist II
Hi BadDad,
Yes, I do have the caps in place, the < 50mV P-P noise are the measurment for it. I too can not find any info in terms of VDD noise level on the data sheet, therefore the question for this thread.
0 Kudos

1,178 Views
bigmac
Specialist III
Hello,
 
I am not sure how you are connecting the resistor in circuit.  However, to drop a 5 volt rail down to 3.6 volts, or less, you will need at least two series connected diodes.  Two diodes would give a drop of 1.4 - 1.6 volts.  If you need to allow that the 5 volt rail can have a maximum tolerance to 5.5 volts, three series diodes would be necessary to not exceed the upper Vdd limit for the MCU.
 
In the absence of a noise specification, the p-p noise applied to the MCU should be significantly less than the internal logic switching level, (which 50 mv p-p clearly is).  However, if you intend using the ADC module within the device, the applied noise level is likely to be far more critical.  With 10-bit resolution, one LSB change will correspond with about 3 mV.  The presence of any noise will give uncertainty to the ADC reading, and may need to be averaged over multiple readings.
 
The alternative to using dropping diodes, is a LDO regulator with 3.3 volt output.
 
Regards,
Mac
 
0 Kudos

1,178 Views
admin
Specialist II
Hi Mac,
I know exactly where you are coming from, in regards to the diode originally I was going for as many that is needed to drop the voltage then use a R to drop the remaining voltage but I ended up using one because I do not have very low R values on hand and a pot
with low R value to trim precisely.

The reason I am worried about the VDD noise is because I will be using ADC, the project is a "precision" multimeter reading V,I and R. So far I am just doing the prototype but the actual project will have the regulator, however for now I need to know if < 50mV noise will effect the QG8.

According to the data sheet the VREFH and VREFL, so it looks like I would need to minimise the noise on VDD as much as possible to get a good ADC reading.

My last option now is to couple an inverting unity gain amplifier to the VDD to kill the ripple. since I have already added common mode inductors to isolate the circuit from the power with appropriate caps.

Any thoughts on that?

Thank you.

0 Kudos

1,178 Views
Encoder
Contributor I
Hi Learner,
 
as BigMac pointed out you cannot even think to use a diode + a resistor to power your 3.3V MCU from 5V supply: the regulation is too much unstable with any variation of the sunk current.
 
I think that 2 or 3 diodes can be enough for using it for logic purpose but you cannot use the ADC for any useful purpose: simply it will not have a stable reference and in any case it will show too much noise.
 
You states that you want to build a "precision" meter with the internal ADC. For that you will need a stable and precise reference with very low noise (<1-2mVpp). A good choice can be a shunt regulator with the ubiquitos TL431 (better in the "A" 1% or the "B" 0.5% grade) for low current requirements or a good and precise 3 terminal LDO regulator like the LP2950/51 (1% typical precision) or things like that.
 
If you need precision you simply cannot avoid to have a precise reference!
0 Kudos

1,178 Views
admin
Specialist II
Hi Enoder,
Thank you so much for your suggestion, I've just wired up the LM732 regulator since thats the only thing available at the local parts shop. However I am still getting < 50mV of 100 Hz ripple from my audjustable output AC adaptor. Can not find the PSRR figure on the QG8, may be the QG8's PSRR is good enough to supress that? Anyone got any ideas on what I should do?

Thanks in advance!


Message Edited by Learner on 2007-12-20 06:20 PM
0 Kudos

1,178 Views
bigmac
Specialist III
Hello,
 
What is the specified worst case drop-out voltage (i.e. the minimum allowable voltage difference between input and output) for the LM732?  It may not be compatible with your 5 volt input, in which case regulation will not occur, and the ripple at the input will be transferred to the output of the regulator.  If this be the case, you will need to somewhat increase your power supply voltage so that proper regulation occurs.
 
A PSRR figure does not make any sense for the ADC of the MCU.  This is because the ADC reference voltage is the voltage difference between Vss and Vdd, so any noise or ripple on these lines will directly influence the conversion accuracy.
 
The use of common mode noise suppression, as you previously indicated, is not appropriate unless the reference point (Vss) of the voltage you are measuring is also downstream of the common mode suppression components.
 
If you are measuring a DC level, it may be possible to compensate for some of the hum component.  This would involve a large number of ADC readings, sampled at a uniform interval, and occurring over the period of an integral number of 50Hz (or 60Hz) cycles.  The reading samples would then be averaged.
 
Regards,
Mac
 
0 Kudos

1,178 Views
peg
Senior Contributor IV
Hi,

I'm not sure what an LM732 is, so I'm guessing you meant LM723.
If this is correct then this device needs at least 3 volts differential between input and output to regulate properly. So you will not get it to work with 5 volts input. You may need a higher tap on your adapter. If this thing is just a transformer watch out for the terrible load regulation of these small,cheap transformers.



Message Edited by peg on 2007-12-21 08:00 AM
0 Kudos

1,178 Views
admin
Specialist II
Hi Peg,
You are right I did mean LM723, the AC adapter for quick convenience I have used the generic multi voltage output adapter. It has selectable 3V, 4.5V, 6V, 7.5V, 9V and 12V output  but the voltage is at least more than 1V off higher.  I suspect the DEMO9S08QG8 is using a similar adaptor for the PSU and it uses a L4931CD33 regulator, I wonder if anyone has such a kit can measure the noise of the VDD on the CRO and make some comment about the noise/ripple?

I am using the Fig.1 circuit of the LM723 data sheet but I have left out Cref according to the LM723 data sheet I should only get about 90uV rms of noise, I am going to put in the Cref and see what happens. It just seems that the regulator is not filtering out the 100hz hum at all, even with common mode inductor at the input and output of the regulator plus bypass caps...


0 Kudos

1,178 Views
peg
Senior Contributor IV
Hi,

Are you sure this is not just  a simple matter of more bulk capacitance prior to the regulator?
Just throw a many thoasand microfarad cap in before the regulator to see if it has any effect.

0 Kudos

1,175 Views
admin
Specialist II
Hi Peg,
I am already using bigger than usual caps like 470uF and 330uF at the input of the circuit with the common mode inductor to filter the adaptor, then the power is further isolated again with another common mode indutor with 330uF cap in front and after before going inot the regulator. The output of the regulator is isolated again with another cap/inductor stage, the caps doesn't seem to have the shorting out effect to the ripple. I think getting a even bigger cap would help but it would need to be fairly rediculously large, if you have a look at the schematic of the DEMO9S08QG8

http://www.freescale.com/files/microcontrollers/hardware_tools/schematics/DEMO9S08QG8SCH.zip?fsrch=1

the biggest filter cap is only 100uF, if my regulator circuit is working correctly which I think it is because it holds the voltage at the adjusted level however seems to be doing a very poor job at rejecting the ripple....

Does anyone have a DEMO9S08QG8 to monitor the noise of the VDD on the CRO? I am certain the ADC of the DEMO9S08QG8 is functioning reasonably well, it would be a good bench mark to see the level of ripple and noise on the DEMO9S08QG8 VDD so I can either try a different regulator until I get rid of the ripple or just let it be and move on with the project.... (already behind in schedule :-(    )

Thanks guys!!

0 Kudos

1,175 Views
bigmac
Specialist III
Hello,
 
Here are the results I obtained for a DEMO9S08QG8 board, when powered from a small 12V plug pack that I had on hand.
 
Ripple at input to regulator ~ 1 volt p-p
Ripple at output of regulator ~ 0.5 mV p-p
High frequency noise at output of regulator ~ 2mV p-p
 
The maximum sensitivity of my oscilloscope is 2mv per division.  The bulk capacitor at the output of the regulator  seems to be 100uF electrolytic.  The demo program is running on the QG8.
 
My understanding of a "common mode inductior" is that it provides coupled inductances in each leg of the supply.  If this is actually the case, I can only suggest that you get rid of it (them), particularly at the output of the regulator.  You require minimal resistance and inductance between the GND terminal of the regulator and the Vss connection of the MCU.  You are actually likely to be measuring the ripple component appearing across the common mode inductor.
 
Regards,
Mac
 
0 Kudos

1,175 Views
admin
Specialist II
Hello friend,
Thank you so much for all your comments, I am very grateful for all your help!

bigmac, your measurement has been a tremendous help! The readings you have indicates that I need to carefully examine my regulator circuit to make sure its fully functional, since my output reading is way off what the data sheet suggests for the part.

In regards to the common inductor and caps filtering, it seems to help to reduce whats coming out of the AC adaptor but has no effect/benifit on the input and the output of the regulator. I will take them out and need to go through the regulator circuit carefully.

Thank you very much my friends!

regards,
0 Kudos

1,175 Views
Encoder
Contributor I
Dear Learner,
the values reported by BigMac are absolutely typical: you do not really needed to measure these values to guess them. What isn't typical is the problems you are experiencing to achieve a faily regulated low noise 3.3 or 3.5V supply with less than 50 or 100mA from an unregulated input which may be some rough 10-14Vdc + 1Vpp ripple (or so).
 
The best solution at all would be a 3 terminal regulator like the cheap LP2950CZ-3.3 or the better LP2950ACZ-3.3 (1.5% max error: didn'you need a good precision and stability?) which may work with 4 to 30V input, needs nothing but 2 capacitors at the input/output terminals (330uF/16V and 100uF/16 are ok), is housed in the handy TO92 package and is self protected.
 
A possible solution if you are not able to achieve this component is to rely on standard ICs and parts. A good one is the cited uA723 which, despite of his old age (it was released by Fairchild in the 1968!) is still capable of delivering well regulated voltage in the range required at the cost of some 6 resistor/capacitors added to the previously 2 for the 3 terminal regulator. This component needs >9.5V input (12 +/-2V is ok) and is capable a 5% very stable output which may be easily trimmed to 0.5% with an added trimmer on the reference input.
 
Another very good solution I used is based on the cheap TL431 + 1 small transistor and a bunch of resistors (4 at all). I used it with precision resistors to achieve <1% error on the output without any trimmer (TL4131A is needed in this case) or with a small added trimmer to slightly adjust the output voltage.
If you need I will be glad to provide some diagrams, provided you states min-max input voltage, input ripple in Vpp, max regulated current (50mA?) and the desired output.
 
Regards from an old boy.
Encoder
 
0 Kudos

1,175 Views
admin
Specialist II
Hi Encoder,
Just out of curiosity, do you have any experience with the uA723 with good results? ie. noise less than 5mV or more? I am looking to minimise the noise on the VDD as much as possible but uncertain if its me or the parts that's not able to achive that......  :smileysad:

Can anyone varify that the 723 regulator can have noise level much less than 10mV? Currently, my AC adaptor is supplying something around 11V into the regulator and putting out 3.5V.

Thanks!
0 Kudos

1,175 Views
Encoder
Contributor I
Dear Learner,
I was a 16 old boy when I bought my first uA723 in a TO100 package in 1968...
I used it for over 30 years whith very good results: in fact it has been a standard component for all these years and still available and cheap today. It is surely capable of 1-2mVpp ripple noise on the output. Its only disadvantage is it is too much complicated for a simple logic.
 
As I pointed out earlier the best guess is a precision 3 terminal regulator. Other ICs I prefer are the TL431/421A/1431 which are far more precise than the old-boy 723 and are cheap and easily available too. As you may see from the schematics I include, the component  number is nearly the same as for the 723.
 
ALL of these circuits should be able to achieve your noise goals, i.e. at least 60-70dB ripple attenuation on the output (<1mVpp-out for 1Vpp-in), provided you make good ground connection at a single star point. Input acceptance is wide for all but the 723 which needs at least 9.5V input: be sure the negative peak of your input ripple does not go lower than this.
 
Good luck and happy Xmas!
Encoder
0 Kudos

1,175 Views
admin
Specialist II
Hi Encoder,
Thank you so much for your generousity and your willingness in sharing your knowledge, I am glad there is someone of your experience around! :smileyvery-happy:

Will looking into the circuits you've shown, thanks again!
0 Kudos

1,175 Views
admin
Specialist II
Hi Encoder,
Thanks for your advise!

I think the part I am using (LM723) is the same as the uA723, I  added the Cref Cap  at the IN+ of fig.1 in the data sheet and that has helped to bring the noise down to < 10mV P-P. It is certainly better but not great, I cant afford to spend much more time on this prototype so I would need to start doing some programming.

Will look into other LDO suggested here when the main frame of the program is done.

I am looking to download Code Warrior Studio, can someone  confirm this would be the one I need to download? 

Special Edition: CodeWarrior for Microcontrollers 
Code Size Restriction:
HC(S)08/RS08 - 32K
ColdFire V1 - 64K


After I hit download,  I can't seem to get pass the form after clicking on Submit. Did anyone ever have the same problem? The browser (firefox) saying loading but has taken a long time and never loads...

Thanks!
0 Kudos

1,175 Views
celsoken
Contributor V
Dear learner,

Just making a slight paradigm shift, maybe you don't need a so stable VDD after all. I've got very good results reading the ADC and the QG8 internal reference in sequence (few useconds). But you'll need a capacitor big enough to keep voltage stable during these two readings' time. The number of samples is up to your software low pass filter resolution/precision scheme. But the quotient ADC reading/Internal reference must be calculated after each reading.

I hope it helps,

Celso


0 Kudos