Why do I have negative dB values after converting PSD output to dB?

I'm using acoustic data which I am in the process of producing spectrograms for. At the moment I am working with one .wav file at a time. I load in the file, apply a calibration correction, perform the spectrogram function and then plot the output of this using surf. In the plotting process, I convert the PSD to decibels. My understanding is that once I do this, all of my values should be positive, is that the case? As, when I check the output of the data, I get a negative value as the minimum value -35.0339. Can anyone explain why this would be??

directory=('Y:\SoundTrap\Noises\Noises_5103_SepOct'); %folder where wav files are

%Load in files of interest:
wavfile='5103.190901150236.wav';

%Get full filename:
wavfilename=fullfile(directory, wavfile);


%Input variables for audioread:
fs=144000;        
tlo=4.0;          %select only part of data and so remove cal.tone
thi=119.0;                      
nlo=fs*tlo;       %multiply fs by tlo to get starting point                    
nup=fs*thi;       %do the same for end point  
    
%Input variables for spectrogram:
nfft=fs;
window=fs;
overlap=window/2;

%Get calibration info:
SN=strsplit(wavfile, '.');
SN1=char(SN(1));
serialNo=str2num(SN1);
               
cal=176.5;    

%Read in file and apply calibration:
        [xbit,fs]=audioread(wavfilename, [nlo nup]); %read in wav 

        xbit=detrend(xbit); %remove DC offset

        cal=power(10,cal/20);   %calculate calibration value
        calxbit=xbit*cal;       %apply calibration
       
        
%Perform FFT and plot:
        f1=figure; %produces empty figure to plot in

        [cal,F,T,P] = spectrogram(calxbit,window,overlap,nfft,fs,'yaxis');  
%P is a matrix representing PSD of each segment

        surf(T,F,10*log10(P),'edgecolor','none');
%plots output of spectrogram
%10log10 converts PSD output to dB

%Look at dB values:
dbvals=10*log10(P); 
mindb=min(dbvals,[],'all');
maxdb=max(dbvals,[],'all');

[Walter Roberson]

Is that -35-ish the value of cal from power(), or the value of calxbit or the value of P or the value of 10*log10(P), or the value of cal from spectrogram() ?

[Louise Wilson]

Sorry for not being clear. The value is in dbvals after I do 10*10log10(P).

[Walter Roberson]

If there is a segment which is fairly quiet (little noise) then it seems to me the PSD for the segment could be small. PSD of roughly 0.00032 would give you the -35 dB

[Louise Wilson]

This is great to know, thank you. I just wanted to be sure it wasn't an error. Thanks!